Conceptio › Archive › arXiv CS

arXiv CSopen access

A multimodal and temporal foundation model for virtual patient representations at healthcare system scale

2026 · arxiv_cs

arXiv CS · Papers · License: Open Access · 2026

Open Source ↗Direct PDF ↓

neural-networks

machine learning, deep learning, neural networks

A multimodal and temporal foundation model for virtual patient representations at healthcare system scale ∗

∗

Andrew Zhang1,2,3,4, , Tong Ding1,2,3,5, , Sophia J. Wagner1,2,3, , Caiwei Tian1,2,3,6 , Ming Y. Lu1,7 , Rowland Pettit1 , Joshua E. Lewis1 , Alexandre Misrahi1,8 , Dandan Mo1 , Long Phi Le1,4,† , and Faisal Mahmood1,2,3,† 1

Department of Pathology, Mass General Brigham, Harvard Medical School, Boston, MA Cancer Program, Broad Institute of Harvard and MIT, Cambridge, MA 3 Data Science Program, Dana-Farber Cancer Institute, Boston, MA 4 Health Sciences and Technology, Harvard-MIT, Cambridge, MA 5 Harvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 6 Department of Biomedical Informatics, Harvard Medical School, Boston, MA 7 Electrical Engineering and Computer Science, Massachusetts Institute of Technology (MIT), Cambridge, MA 8 School of Computer and Communication Sciences, EPFL, Lausanne, Switzerland ∗ Contributed equally (Co-first) † Co-senior authors

arXiv:2604.18570v1 [cs.LG] 20 Apr 2026

Abstract Modern medicine generates vast multimodal data across siloed systems, yet no existing model integrates the full breadth and temporal depth of the clinical record into a unified patient representation. In the present study, we describe a comprehensive effort to model human clinical trajectories. We introduce A POLLO, a multimodal temporal foundation model trained and evaluated on over three decades of longitudinal hospital records from a major US hospital system, composed of 25 billion records from 7.2 million patients, representing 28 distinct medical modalities and 12 major medical specialties. A POLLO learns a unified representation space integrating over 100 thousand unique medical events in our clinical vocabulary as well as images and clinical text. This “atlas of medical concepts” forms a computational substrate for modeling entire patient care journeys comprised of sequences of structured and unstructured events, which are compressed by A POLLO into virtual patient representations. To assess the potential of these whole-patient representations, we created 322 retrieval and prognosis tasks from a held-out test set of 1.4 million patients. We demonstrate the generalized clinical forecasting potential of A POLLO embeddings, including predicting new disease onset risk up to five years in advance (95 tasks), disease progression (78 tasks), treatment response (59 tasks), risk of treatment-related adverse events (17 tasks), and hospital operations endpoints (12 tasks). Linear probing with A POLLO embeddings exhibits strong performance across all task categories. On a 30-task cancer progression benchmark, A POLLO surpasses even task-specific fully-supervised transformer models. Using feature attribution techniques, we show that model predictions align with clinically-interpretable multimodal biomarkers. We evaluate semantic similarity search on 61 retrieval tasks, and moreover demonstrate the potential of A POLLO as a multimodal medical search engine using text and image queries. Together, these modeling capabilities establish the foundation for computable medicine, where the full context of patient care becomes accessible to computational reasoning. 1

Introduction Electronic Health Records (EHRs) function as the system of record for modern medicine, underpinning the tightly coupled domains of care, operations, and research by systematically capturing complex, longitudinal health trajectories. These rich repositories hold immense promise for precision medicine and the creation of a unified, continuously learning healthcare system [1, 2]. Yet, healthcare faces a paradox: while generating nearly 30% of global data volume at 50 petabytes annually, less than 3% is effectively utilized for clinical insight [3, 4]. The imperative to close this gap is underscored by the devastating impact of fragmented care, which contributes to an estimated 795,000 diagnostic error-related deaths or permanent disabilities annually in the United States alone [5]. Unlocking the full potential of EHRs is currently hindered not only by systemic fragmentation across modality-specific silos, but also inherent data complexities – extreme dimensionality, irregularity, and sparsity [6, 7]. In recent years, foundation models have emerged as a paradigm shift across domains such as natural language processing [8, 9], computer vision [10, 11] and computational biology [12, 13], learning context-rich representations via large-scale self-supervision that transfer effectively to diverse downstream tasks [14]. While their potential in healthcare is recognized [15, 16], existing models remain limited in scope. Most approaches analyze individual data modalities in isolation – such as pathology images or clinical text – which inherently limits their utility for tasks requiring a holistic view of the patient [17–24]. Efforts to build longitudinal EHR foundation models [25–35] have largely been restricted to structured data (e.g., diagnostic codes or laboratory results) [36] due to the ease of leveraging existing ontologies (e.g., ICD-10, SNOMED, LOINC), data warehousing infrastructure, and discrete-token sequence modeling techniques inspired by large language models (LLMs). However, structured data provides only a limited snapshot of a patient; subtle nuances such as medical reasoning traces, disease progression, and novel biomarkers are often only available in free-text notes and imaging. With recent evidence suggesting that multimodal AI systems can outperform single-modality approaches by 6–33% across diagnostic tasks [37, 38], there is a critical need for unified models capable of synthesizing the full spectrum of patient data. This challenge mirrors recent efforts in creating virtual cells, where foundation models that integrate multimodal data – for example, synthesizing transcriptomic, proteomic, and morphological signals to predict cellular responses [39, 40] – have demonstrated how unified representations can capture complex system dynamics that single-modality approaches miss. Just as these models revealed emergent properties in cellular state transitions, virtual patients could unlock similar insights into health trajectories by synthesizing the full spectrum of longitudinal EHR data. To realize this vision, we introduce A POLLO, a multimodal temporal foundation model that transforms the entirety of the longitudinal medical record into a unified computational representation. A POLLO first tokenizes individual medical events using modality-specific encoders and then fuses the resulting embeddings across timepoints using a transformer architecture (see Results and Methods for full architectural details). We develop A POLLO with over 25 billion medical events from 7.2 million patients spanning 33 years across a multi-institutional healthcare system, encompassing text-based, structured, and imaging modalities. Unlike curated research datasets limited to single departments or critical care cohorts [41], our corpus captures the 2

full continuum of inpatient and outpatient care across 17 institutions and affiliated clinics, reflecting real-world heterogeneity in both practice and population. We evaluate A POLLO across 322 downstream tasks designed to assess clinical breadth, including patient retrieval (61 cohorts), predicting new disease onset (95 tasks), disease progression (78 tasks), treatment response (59 tasks), drug adverse events (17 tasks), and hospital operations endpoints (12 tasks). The whole-patient embeddings achieve strong performance on prognostic tasks, enable multimodal guided retrieval from a large-scale database, yield interpretable latent structures that mirror phenotypic clusters, and reveal multimodal biomarkers. Together, these results demonstrate that unified, temporally grounded patient representations can convert the EHR from a static archive into an active computational substrate for care, research, and operations – the foundation for next-generation computable medicine.

Results Building a foundation model for clinical time series data A POLLO is a transformer-based model that integrates the entirety of a longitudinal, multimodal medical record, including clinical reports, structured data, and imaging data, in a temporal context. We curated MGB-7M, consisting of 7,155,044 patient electronic health records (EHR) with 25,296,943,893 distinct medical events. MGB-7M comes from a multi-institutional healthcare organization including specialty, ambulatory, and teaching hospitals (Figure 1a). The events span 33 years of which the majority of events are in the past 15 years. The dataset is roughly balanced between male (3,915,625) and female (3,236,747) patients, with the majority of patients between 18 and 79 years old at the time of their last medical record (Extended Data Table 1). MGB-7M covers the entire diagnostic spectrum including flowsheet measurements (22,936,351,260 events), laboratory tests (1,442,631,333 events), vital signs (505,193,723 events), and text-based observations including progress notes (158,683,290 notes) and diagnostic reports from various medical domains (77,978,881 reports), and imaging data (1,158,235 images) including hematology and histopathology imaging (Figure 1b-e, Extended Data Table 2). MGB-7M covers a diverse spectrum of clinical entities with diagnostic reports from 12 major medical domains (Extended Data Table 2), medications covering all 14 chapters of the Anatomical Therapeutic Chemical (ATC) Classification (Extended Data Table 3), and disease codes spanning all chapters of the International Classification of Diseases (ICD) ontology [42], with on average 1.8 million diagnoses per chapter (Extended Data Table 4). A POLLO integrates all medical events in the patient history into a unified, temporally aware mathematical representation. Inspired by large language models, we convert the medical record into a sequence of tokens: structured data are mapped to sequences of discrete event tokens, while unstructured data modalities are read as text tokens or image patch tokens. Each modality – text, imaging, or structured – is independently converted into continuous embeddings (value tokens) using modality-specific adapters. More specifically, we encode text-based modalities using a clinical large language model [18], we encode images using medical vision foundation models [43, 44], and we map structured tokens to learnable embeddings (Figure 1f). Subsequently, all tokens are projected with modality-specific layers to a common representation space which reflects the seman3

Measurements

.9B

.4B

22 1

Flo

sheet

Lab

ody

measurements Physical Pulmonary

respiratory

hemistry

7.2M patients

ardiac

ematology

Device

Administrative

Drug

3 3B 3 3B 3 1B / 2 9B

linical report

health system

toxicology

ther

Progress note

ount

Multi-institutional

Observations

Medications

Systemic

ount (log scale)

hormones

Antiparasitic

Genitourinary

Radiology

imaging

954 700 627 518 472 434 1 5B

system

Alimentary and

Pathology

ardiology

metabolism

ervous system

ardiovascular

phthalmology

system

Vascular studies

Dermatologicals

uclear medicine

Endocrinology

Various Respiratory system

Antineoplastic

Interventional

agents

procedures

lood-related

Laboratory

Molecular

Antiinfectives

genetics

k O

Sensory organs

eurology

Musculo-s

Pulmonary

25B record entries

Ethnicity

Sex

MLP

Image

g Age

Multimodal encodings

VFM

ther

Visit

C 0 5 10 15 566K 328K 242K 235K 182K 167K 57 3K 30 2K 17 1K 9 8K 6 8K 5 6K 5 3K 64 110

ount (log scale)

0 4 8 12 16 & 46 6 12 9 N 84 C C 19 O 15 11 N 961K 829K B 579K / 276K 214K N 130K O 24

iagnosis

C 16 0 2 4 55 fi 54 k & 33 75 C & 22 94 21 19 18 N 16 15 N 15 14 k 884K & 699K 668K B & 573K 440K 247K C 243K 132K 128K ount

ther clinical note

1 1M Images

D 39 ICD10 codes

112M Drug components

Diagnostic reports

5M Vitals

78M

4M O C 19 1 2 3 .

w s

eletal

system ther

H C

ealth services

linical

Digestive

Genitourinary

Respiratory

ndings

Musculos

eletal

Endocrine

metabolic irculatory Mental

behavioral

ervous system

Injury

Eye

poisoning Infectious

lood

immune Ear

Pregnancy

Perinatal

ongenital

Special purposes External causes

Visit n Value to

ens

fi k

(modality-speci

k k

Mas +

MLP

LLM

“...”

eoplasms

Time to Text

ed to

ens (in age)

APOLLO encoder

H Emb.

data

MLP

idden state

Structured

APOLLO decoder

(<id>,<value>) Reconstruction

h Patient embeddings

Time-to-event tasks

Insert prompt, at time tk

Patient retrieval

τ Patient embedding

at time step k

Discharge

Disease

APOLLO

EMR

Patient

k-1

Disease

2 2

APOLLO Disease onset

Disease progression

MED

APOLLO

from patient Med

Treatment efficacy

database

Demographics

Figure 1: Overview of MGB-7M and A POLLO. (a) Overview of the pretraining dataset MGB-7M curated from 17 hospitals in one large-scale health care system consisting of 7.15 million patients. (b-e) Detailed distribution of MGB-7M including (b) LOINC code distribution of measurements, (c) distribution of diagnostic reports across medical domains, (d) medications grouped by ATC classification, and (e) ICD10 codes grouped by ICD10 chapters. (f) Modality-specific encodings of imaging data, text-based clinical notes and diagnostic reports, and structured measurements such as laboratory tests, medications, and diagnosis. (g) Model architecture of A POLLO, an encoder-decoder architecture taking patient demographics (age, sex, ethnicity) and embeddings of all patient records in a temporal context as input. The pretraining objective is masked modeling, where modality-specific masked tokens are reconstructed. (h) At inference, all patient records are fed into the model together with one additional token that yields the patient embedding at the last time step. (i) Schematic of the curation of time-to-event tasks for risk prediction of disease onset, disease progression, and treatment outcome. (j) Schematic for patient retrieval for a query patient embedding (blue) from a large-scale database of candidate patients (gray).

tic relationships between events. We prepend patient demographics (age at last event, sex, and ethnicity) to the input sequence and add the per-timestep patient age to each event as a positional (time) encoding (Figure 1g). By isolating the main computational unit (the temporal transformer) from the raw data, we reduce the risk of protected health information (PHI) leakage since no raw data enters the final model. A POLLO effectively scales training to 28 modalities from an active multi-institutional EHR system across all departments, satellite clinics, covering both inpatient and outpatient visits. This is in contrast to existing EHR modeling efforts [31, 32, 45] 4

utilizing curated, publicly accessible datasets which have undergone substantial curation and preprocessing, are restricted to specific clinical departments such as MIMIC-IV [41], or are limited to structured data modalities only [46]. We pretrain A POLLO employing masked token modeling [9] with modality-specific masks that prompt the output head for the correct modality during the forward pass (for more details see Section Pretraining in Online Methods, Extended Data Table 5). To generate a patient embedding during inference, the last (most recent) time token is appended to the input sequence along with a modality-specific mask token as prompt, and the corresponding hidden state is used as the patient embedding (Figure 1h). In practice, any modality mask can be used as the prompt, and the resulting patient embedding is tuned towards that modality. For example, we later explore the applications of text- and image-prompted patient embeddings in multimodal semantic retrieval. For the majority of downstream tasks, however, we utilize the diagnosis mask as the prompt, due to the disease-oriented nature of most tasks. We evaluate A POLLO on a wide set of 261 prognostic and 61 retrieval tasks designed to demonstrate the general-purpose utility of these patient-level embeddings. To assess prognostic value, we train Cox proportional hazards models for time-to-event (TTE) prediction using the frozen A POLLO patient embedding as features (Figure 1i). Clinical endpoints include new disease onset, disease progression, treatment response endpoints, drug-related adverse events, and acute hospital operations endpoints. Beyond prognosis, A POLLO patient embeddings can be used as a search index for retrieving patients with similar disease history and medical phenotypes. Lastly, we extract insights into the clinical factors driving model risk predictions, marking a step towards biomedical discovery across the entirety of the medical record (Figure 1j).

A POLLO’s embedding space as an atlas of medical concepts For the first time, A POLLO brings siloed events across the entirety of the medical record into a unified embedding space, permitting the detailed study of intra- and cross-modal relationships. To understand how A POLLO unifies data representations across modalities, we examine the input embedding space of A POLLO after pretraining on MGB-7M. The structured modalities, i.e., flowsheets, vitals, laboratory tests, medications, and diagnosis codes, generalize across medical events and, by analogy to language modeling, form A POLLO’s vocabulary used to compose patient timelines. The low-dimensional representations of all 103,940 discrete tokens that occurred more than 100 times in the records show that A POLLO captures similar medical concepts by grouping flowsheets, vitals, and lab tests into distinct semantic clusters, such as allergy, coagulation, drug & toxicology, or hematology, aligning with parent categories in the LOINC ontology (Figure 2a). Using modality-specific masks during pretraining, A POLLO learns a natural separation between modalities, such as diagnostic codes and medications. Upon closer examination of the diagnostic tokens, we observe that the disease tokens of related diseases are closely embedded (Figure 2b). Across the ICD10 chapters, diseases of the eye cluster together with eye-related diseases from other chapters, such as benign neoplasms of eye and adnexa. Within diseases of the circulatory system, both acquired and congenital heart-related diseases are neighboring acute myocardial infarction. Similarly, different lymphomas cluster closely together. These results show that A POLLO learns multiscale semantic relationships between underlying conditions and 5

a Structured modalities

(flowsheet, vitals, labs, diagnosis, medications)

b Diagnosis (ICD10 chapters)

Eye

D31: Benign neoplasm of eye and adnexa Q14: Congenital malformation of posterior segment of eye H13: Chorioretinal inflamation T26: Burn and corrosion confined to eye and adnexa

Circulatory Q25: Congenital malformation of greater arteries I21: Acute myocardial infarction I42: Cardiomyopathy I44: Angina pectoris

C83: Non-follicular lymphoma C82: Follicular lymphoma C81: Hodgkin lyphoma

Lymphoma c Medications

Immuno-

therapy

Pembrolizumab Nivolumab Ipilimumab

Type 2 Diabetes

Glipizide Metformin Insulin types Glucose

HIV

The overall morphologic findings, in the setting of significant leukocytosis, are suspicious for chronic myeloid leukemia (CML).

Serum urate: 8.1-9.9 mg/dl Serum lactate dehydrogenase: 796.9-1050 u/l Albumin in serum or plasma: 4.49-4.9 g/dl

Cytogenetic analysis of an unstimulated bone marrow specimen demonstrated a translocation involving two chromosomes in all analyzed metaphases, corresponding to the BCR::ABL1 fusion. This finding is characteristic of Chronic Myeloid Leukemia (CML) and a subset of cases of Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML).

No pathogenic single nucleotide variants or small insertions/deletions were detected.

Bone marrow biopsy Bone marrow aspirate Cytogenetics

Hematology

Blood smear

Bathing: independent Feeding: independent

Coagulation Cytology Device Diagnosis Drug & Toxicology

Blood smear (CellaVision images)

WBCs [#/volume] in blood by automated count: >19.9 k/ul Metamyelocytes/100 WBCs in blood by manual count: >2.64 k/ul

Neutrophils/100 WBCs in blood by automated count: >90.30%

Systolic blood pressure: 123.8-134.8

Progress note 1

Persistent hypotension MAP: 0-0.09

Walk in home: independent

Cardiac Cell Markers Chemistry Clinical Report Clinical Trials

Lymphocutes [#/volume] in blood: >4.03 k/ul

Platelets [#/volume] in blood by automated count: 147.8-198.5 k/ul

Progress note 2 Sequencing panel

Dressing: independent

High risk screen: chronic illness Administrative Allergy Antibiotic Susceptibility Blood Bank Body Measurements

Hematology

Sequencing panel findings

Cytogenetic findings

Alimentary tract and metabolism Antiinfective for systemic use Antineoplastic and immunomodulating Antiparasitic, insecticides, and repellents Blood and blood forming organs Cardiovascular system Dermatologicals Genito urinary system and sex hormones Musculo-skeletal system Nervous system Respiratory system Sensory organs Systemic hormonal preparations Unknown Various

30-40 y.o., male, suspected CML

d Unstructured modalities Bone marrow biopsy findings

Atazanavir Abacavir / lamivudine Rilpivirine Lamivudine

1: Infectious diseases 2: Neoplasms 3: Blood & immune 4: Metabolic 5: Mental & behavioral 6: Nervous system 7: Eye 8: Ear 9: Circulatory 10: Respiratory 11: Digestive 12: Skin 13: Muscoskeletal 14: Genitourinary 15: Pregnancy 16: Perinatal 17: Congenital 18: Clinical findings 19: Injury & poisoning 20: External causes 21: Health services 22: Special purposes Death

Mean arterial pressure: 93-101.8 Administrative Emergency Fertility Gastrointestinal Genetics Hematology

History Input/Output Laboratory Medication Microbiology

Miscellaneous Molecular Obstetrics/Gynecology Organ Transplant Pathology

...

Progress note 1 The patient presented to the emergency department with leukocytosis. Prior evaluation by a primary care provider, including routine laboratory testing, demonstrated a markedly elevated white blood cell count (>70,000) with a left-shifted differential. Physical Physical Therapy Pulmonary/Respiratory Serology Specimen

Urinalysis Vaccination

Figure 2: A POLLO generates an atlas of medical concepts. (a) Uniform manifold approximation and projection (UMAP) of the 103,940 discrete tokens that occur more than 100 times shows that A POLLO learns the underlying semantics of the discrete concepts. The emerging atlas of medical concepts exhibits meaningful spatial relationships both within modalities, as seen for (b) diagnosis codes and (c) for medications, as well as across modalities (d). (d) Structured and unstructured embeddings of the health record entries of a visit of one patient (30-40 y.o., male, suspected chronic myeloid leukemia), projected to the UMAP fit of (a).

associated anatomical organs that align with but are not limited by clinician-defined ontologies (e.g., ICD-10). The embeddings of prescribed medications and drug components labeled by RxNorm ingredient code show similar semantic clustering (Figure 2c). In particular, A POLLO closely embeds drugs such as pembrolizumab and nivolumab, both used in immunotherapies targeting the PD-1 protein in cancer cells, and antiretroviral drugs, such as abacavir, lamivudine, or rilpivirine, which are used in combination to treat HIV infections. Drugs related to type 2 diabetes, such as glipizide and metformin, are grouped together in close proximity to blood sugar indicators insulin and glucose. Beyond discrete tokens encoding structured information, A POLLO additionally encodes event-specific diag-

nostic reports, progress notes, or clinical images. Remarkably, these unstructured event embeddings map to locations in the atlas which correspond to semantically related structured tokens. To understand how text and images integrate with the discrete input embedding space, we analyze all embeddings of one patient (30-40 y.o., male) in one day of their medical record, consisting of 184 lab tests, 58 flow sheet entries, one medication, six clinical reports, three progress notes, and images from one blood smear (Figure 2d). We observe that the clinical reports from the bone marrow aspirate and bone marrow biopsy are embedded close to the cytogenetic report, all three situated within the hematology cluster in green. All blood count-related lab tests are located in the same hematology cluster as are the blood smear images. The closest lab tests to the blood smear show elevated levels of neutrophils and their precursor form metamyelocytes, in line with the observation of leukocytosis mentioned in the biopsy note and progress note. The progress note describing the patient visit is located in the administrative events cluster together with routine flowsheet entries. In a completely self-supervised manner, A POLLO learns clinician-aligned representations of diverse medical events, both structured and unstructured, forming a multimodal atlas of medicine. These embeddings are the building blocks for modeling representations of the patient state.

A POLLO patient embeddings encode medical phenotypes Through the time-aware transformer architecture, the individual medical events encoded by A POLLO are composed into patient-level representations which encode the entire clinical history of a patient. In a UMAP of 100 thousand randomly sampled patients, each at a random timepoint, we observe that their embeddings reflect underlying disease phenotypes, age gradients, and shared medical histories (Figure 3a). We highlight clusters related to breast cancer, pregnancy, blood cancers, skin disorders, and metabolic disorders. Just like the individual event embeddings, these clusters are learned from scratch in a self-supervised manner. Because patient embeddings are dynamic and change over time, we can trace their trajectories as patients approach the onset of a disease or experience disease progression. For a cohort of patients who are eventually diagnosed with schizophrenia, their embeddings converge to the same region of the embedding space as they approach the diagnosis date (red stars; Figure 3b). Similarly, patient trajectories of a cohort in the months following a diagnosis of glaucoma diverge into a group eventually developing blindness and a group showing different mostly unrelated conditions (Figure 3c).

A POLLO predicts risk of new disease onset We evaluate the predictive capabilities of A POLLO on 95 downstream tasks for new disease onset. More specifically, we curate endpoints including 94 distinct disease codes spanning eight categories ranging from cardiovascular and circulatory diseases to cancers, as well as all-cause mortality (Figure 3d, Extended Data Table 6). Each disease onset task includes at least 60 thousand patients in total, where the number of uncensored patients (those who experience the endpoint) ranges from 157 to 27,503 per task reflecting highly variable real-world disease incidences. All tasks are curated from the test set of MGB-7M spanning 1,431,009 patients in total with a varying number n of patients considered per task depending on the inclusion criteria.

b Schizophrenia

31F, normal first pregnancy

+/PR-/

57F, high grade ER

+ breast cancer

HER2

treated with chemo,

70F, metastatic breast cancer to lung

G34.108:

37F, primary female

with aura

Skin disorders

Migraine

induced psychotic disorder

72M, seborrheic keratosis of upper back

35F, acute myeloid

Disease progression

disease

d New onset 0.65

All-Cause Mortality

0.50

0.92

0.71 0.64

Gastrointestinal, Hepatic & Renal (N=12) 0.60

0.55

Cardiovascular & Circulatory (N=11)

0.80

Respiratory & Infectious (N=7)

0.78

Oncology (Cancers) (N=33)

0.77

Endocrine & Reproductive (N=11)

0.59

0.76

Musculoskeletal & Autoimmune (N=4)

0.59

0.67 0.66

Hematology & Sensory (N=10)

0.63

0.5

0.6

0.7

Heart Failure (n=145k,τ≤3.0y,I=7.7%)

0.8

Acute Myocardial Infarction (n=153k,τ≤1.0y,I=0.72%)

0.67

Intracranial Hemorrhage (n=198k,τ≤1.0y,I=0.24%)

0.71

0.9

****

0.77

0.61

0.5

0.6

Hypertension to Heart Failure (n=138k,τ≤5.0y,I=4.0%)

1.0

0.88

AUROC 0.7 0.8

Schizophrenia (n=199k,τ≤1.0y,I=0.1%)

0.65

Bipolar Disorder (n=194k,τ≤1.0y,I=0.3%)

****

0.86

****

0.83

0.92 Asthma to Exacerbation Requiring Oral Corticosteroids (n=58k,τ≤3.0y,I=2.8%) 0.86 COPD to Exacerbation

****

0.63 0.59

****

0.75

COVID-19 Mortality (n=14k,τ≤60d,I=0.20%)

****

(n=12k,τ≤3.0y,I=3.0%)

0.86

0.82

(n=167k,τ≤3.0y,I=0.36%)

0.78

Type 2 Diabetes (n=166k,τ≤3.0y,I=7.7%)

0.61

Gout (n=170k,τ≤1.0y,I=0.75%)

****

0.74

0.85 0.85

****

0.83

g Adverse events

****

0.71

Prostate Cancer Mortality (n=13k,τ≤5.0y,I=15.6%)

0.74 0.75

(n=2k,τ≤5.0y,I=19.7%) AUROC 0.6

0.7

0.8

0.9

1.0

****

0.69

****

0.67

Exacerbation-Free Survival on Tiotropium for COPD (n=2k,τ≤3.0y,I=45.5%)

0.59

Exacerbation-Free Survival on Umeclidinium / Vilanterol for COPD (n=1k, 3.0y,I=53.1%)

0.60

0.75

0.71

Top4 Oncology (Cancers) Survival on Trastuzumab for Breast Cancer (n=4k,τ≤3.0y,I=14.8%)

****

0.66

Survival on Pertuzumab for Breast Cancer (n=1k,τ≤3.0y,I=7.1%)

0.87 0.84

Survival on Paclitaxel for Breast Cancer (n=1k,τ≤3.0y,I=21.4%)

0.83

Survival on Leuprolide for Prostate Cancer (n=2k,τ≤2.0y,I=12.3%)

h Hospital operations

0.6

0.7

0.8

GI Bleeding on NSAIDs (n=134k,τ≤90d,I=0.10%)

0.80

0.91

Acute Catheter-Associated UTI (n=90k,τ≤6d,I=0.01%)

0.82

Acute Pressure Ulcer (n=105k,τ≤6d,I=0.07%)

0.70

Angioedema on ACE Inhibitors (n=38k,τ≤90d,I=0.05%)

*** *

0.60

GI Bleeding on Antiplatelet Agents (n=55k,τ≤1.0y,I=0.85%)

0.69

Cardiomyopathy After Anthracyclines (n=6k,τ≤5.0y,I=9.3%) *

0.54

Hyperkalemia on ACE Inhibitors (n=20k,τ≤30d,I=0.41%)

0.74

Acute Venous Thromboembolism (n=130k,τ≤6d,I=0.23%) Acute Acute Liver Failure (n=129k,τ≤6d,I=0.01%) Inpatient Readmission (All Cause) (n=41k,τ≤30d,I=27.3%)

0.54

Acute ARDS (n=276k,τ≤6d,I=0.01%)

0.54

0.67

0.71

Colorectal Cancer

Thyroid Cancer

0.74

0.56

0.72 **** ****

0.62

0.57

Breast Cancer

Bladder Cancer

Mortality

Leukemia Mortality

Mortality

Metastasis

Mortality

Metastasis

Palliative Care

Mortality

(n=13k(2k),τ≤5y)

(n=2k(442),τ≤5y)

(n=18k(3k),τ≤3y)

(n=4k(1k),τ≤3y)

(n=7k(706),τ≤3y)

(n=6k(2k),τ≤3y)

(n=6k(463),τ≤5y)

(n=14k(596),τ≤5y)

(n=4k(877),τ≤3y)

0.78

Bladder Cancer

0.79 0.76

0.79

All Leukemia

**** ****

0.59

0.74

0.80

****

0.66 0.63

0.73

0.80

0.68

Acute Sepsis (n=124k,τ≤6d,I=0.10%) Inpatient Length of Stay (n=138k,τ≤6d,I=81.9%)

Breast Cancer

0.83 ****

0.64

0.79

1.00

Acute Clostridioides Difficile Infection (n=136k,τ≤6d,I=0.05%)

0.77

0.88

0.58

0.83

(n=5k(789),τ≤3y)

0.82 0.80

0.67

0.97 0.91

0.83

Chronic Myeloid

Acute Cardiac Arrest (n=141k,τ≤6d,I=0.01%) Acute Acute Kidney Failure (n=114k,τ≤6d,I=0.51%)

1.0

0.84

Prostate Cancer

0.84 0.83

0.9

****

0.65

****

0.58

MACE on Levothyroxine with History of Heart Failure (n=2k,τ≤2.0y,I=36.5%)

Melanoma Mortality

****

0.71

Heparin-Induced Thrombocytopenia (n=142k,τ≤14d,I=0.01%)

Clostridioides difficile Infection on PPIs (n=45k,τ≤1.0y,I=0.49%)

0.87

****

0.69

0.86

BACC

0.5

Acute Dialysis Dependence (n=108k,τ≤6d,I=0.02%)

0.55

0.88

****

0.91

Tardive Dyskinesia on Antipsychotics (n=82k,τ≤2.0y,I=0.07%)

0.89

0.75 0.65

****

Type 2 Diabetes After Atypical Antipsychotics (n=16k,τ≤2.0y,I=3.0%)

0.93

0.75

0.80

Type 2 Diabetes After Systemic Corticosteroids (n=135k,τ≤2.0y,I=1.9%)

0.79

****

Acute Kidney Injury After NSAIDs (n=134k,τ≤30d,I=0.25%)

****

1.0

0.80

****

0.63

0.9 0.81

Respiratory & Infectious

0.79

Chronic Myeloid Leukemia Mortality

0.5

0.73

AUROC 0.7 0.8

0.6

MACE-Free Survival on Losartan for Hypertension (n=8k,τ≤3.0y,I=22.4%)

0.83

****

0.5

MACE-Free Survival on Amlodipine for Hypertension (n=15k,τ≤3.0y,I=27.0%)

0.87

****

0.57 0.57

Melanoma Mortality (n=5k,τ≤3.0y,I=8.7%)

Hyperprolactinemia

0.68 0.66

Apixaban for Atrial Fibrillation

(n=3k, 3.0y,I=35.9%)

Top3 Oncology (Cancers)

Top3 Endocrine & Reproductive

0.69

0.62

Stroke/Embolism-Free Survival on

1.0

0.72

Aortic Stenosis Mortality (n=7k,τ≤3.0y,I=14.9%) Top3 Respiratory & Infectious

0.59

Top3 Cardiovascular & Circulatory 0.9

0.84

0.82

Oncology (Cancers) (N=36)

0.75 0.73

0.63

****

0.81

0.59

Neurology & Psychiatry (N=1)

0.72

****

0.80 0.64

Respiratory & Infectious (N=2)

Gastrointestinal, Hepatic & Renal (N=2)

Stroke Mortality (n=16k,τ≤3.0y,I=14.3%)

****

Top3 Neurology & Psychiatry

Epilepsy (n=217k,τ≤1.0y,I=0.4%)

0.72 0.70 0.68

0.72 Top3 Cardiovascular & Circulatory

Endocrine & Reproductive (N=1) 0.57

Cardiovascular & Circulatory (N=17)

0.68

Neurology & Psychiatry (N=3)

AUROC

Top3 Cardiovascular & Circulatory

0.75 0.75

Hematology & Sensory (N=1) 0.53

0.75 0.74

0.55 0.60 0.65 0.70 0.75 0.80

0.74

0.60

0.64 0.63

AUROC

0.75

0.65

Respiratory & Infectious (N=6)

Oncology (Cancers) (N=22)

0.70

0.63

Endocrine & Reproductive (N=12)

Musculoskeletal & Autoimmune (N=9)

f Treatment efficacy

0.65

0.62

Gastrointestinal, Hepatic & Renal (N=8)

0.78

0.60

E78 Hyperkalemia

AUROC

0.95

0.72

Cardiovascular & Circulatory (N=11) Neurology & Psychiatry (N=12)

0.85

Colon polyp

renal failure

rolemia & arthritis

e Disease progression

0.75

End stage

81M, hypercholeste-

AUROC 0.55

Osteoporosis

2 diabetes, chronic kidney

50 Age

Blindness within 2 years

infection (UTI)

63M, hyperlipidemia, type

Urinary tract

disease, severe obesity

transplant

No Blindness in 2 years Blindness diagnosis

Metabolic disorders 59M, chronic kidney

peripheral blood cell

Schizophrenia diagnosis

Schizophrenia in 1 year

c Glaucoma

plaque on right cheek

leukemia s/p chemo &

All patients

65F, hyperkeratotic

Blood cancers

0.75

Apollo

Apollo (structured-only)

Supervised (Apollo w/o pretraining)

Last-progress-note

0.63

0.64

0.73

0.76

0.49

0.61

0.60

0.69

0.76

0.65

0.61

0.71

0.74

0.77

0.54

0.60

0.65

0.75

0.57

0.61

0.67

0.73

0.66

0.62

0.74

0.77

0.63

0.72

0.77

0.75

0.52

0.60

0.80

0.74

0.73

0.76

0.83

0.71

0.67

0.00

0.73

0.25

0.80

0.50 0.87

AUROC

unspecified with inhalant-

of cheek & forearm

prior thyroid nodule biopsy

F18.959: Inhalant use,

81M, basal cell carcinoma

72F, multiple myeloma with

Onset Fluency Disorder

pregnancy

infertility & ovarian cysts

radiation, surgery

F80.81: Childhood-

New onset

21F, contractions during

+ 1) (days)

Pregnancy

Breast cancer

38F, stage III triple negative breast cancer

log(TTE

Age-sex

Figure 3: Evaluation of A POLLO’s patient embeddings. (a) Uniform Manifold Approximation and Projection (UMAP) visualization of 100 thousand randomly sampled patient embeddings from the data partition for downstream evaluation, labeled by age. Local neighborhoods reveal clustering of patients with similar clinical phenotypes. (b) Patient trajectories of 10 random patients before their diagnosis of Schizophrenia shows convergence to the same cluster in the patient embedding space. (c) Patient trajectories of 10 random patients with Glaucoma, five of which develop blindness within two years. The patient trajectories diverge from the disease cluster, where the patients with later diagnosed blindness cluster together. (d-h) Quantitative evaluation on 261 curated downstream tasks. Performance of A POLLO (blue) and the age–sex reference (orange) is given in AUROC scores across all tasks. (d) New disease onset prediction (95 tasks), (e) disease progression prediction (78 tasks), (f) treatment response prediction (59 tasks), (g) adverse effect (17 tasks, where top 12 tasks are shown in the figure), and (h) hospital operation tasks (12 tasks). (i) Ablation study on architectural design choices of A POLLO, evaluated on disease progression of neoplastic diseases. Top 10 tasks are shown, more tasks can be found in Extended Data Figure 17. Tasks in each plot are sorted by A POLLO’s performance. Statistical significance was assessed by an unpaired, nonparametric bootstrap test of the baseline with respect to A POLLO. ∗ p ≤ 0.05, ∗∗ p ≤ 0.01, ∗∗∗ p ≤ 0.001, ∗∗∗∗ p ≤ 0.0001. N, number of tasks; n, number of patients; τ , time-to-event; I, class imbalance in test set.

For each disease, we predict time to diagnosis starting from a random hospital discharge (hereafter referred to as the snapshot time; see Evaluation Framework in Online Methods) using a Cox-regression model on PCA-transformed A POLLO patient embeddings. As the primary performance metric, we report time-dependent cumulative/dynamic AUC evaluated at a fixed threshold duration between one and six years in the future depending on the physiological time course of the disease. As a statistical reference, we train separate Cox models using patient age and sex as features. Overall, A POLLO significantly outperforms the baseline on 74 out of 95 diseases (p < 0.05). Per disease category, A POLLO performs best on risk prediction of all-cause mortality within the next year (0.92 AUROC). Investigating the specific diseases in the best-performing categories, we find that A POLLO significantly improves risk prediction over age-sex features for cardiovascular and circulatory diseases: 3-year heart failure prediction improves from 0.77 to 0.88 AUROC, 1-year acute myocardial infarction improves from 0.68 to 0.82 AUROC, and 1-year intracranial hemorrhage improves from 0.71 to 0.81 (all p < 0.0001). For conditions related to neurology and psychiatry, A POLLO predicts 1-year risks for schizophrenia with 0.92 AUROC compared to 0.65, bipolar disorder with 0.86 AUROC compared to 0.63, and epilepsy with 0.86 AUROC compared to 0.59 (all p < 0.0001). For diseases of the endocrine and reproductive system, A POLLO significantly improves the 3-year risk prediction of hyperprolactinemia from 0.78 to 0.85 (p < 0.01), for 3-year risk of type 2 diabetes from 0.61 to 0.85 (p < 0.0001), and for 1-year risk of gout from 0.74 to 0.83 (p < 0.01) AUROC, respectively (Figure 3d). Results for all experiments can be found in Extended Data Table 7.

A POLLO patient representations predict disease progression To assess the performance of A POLLO in predicting disease progression in more targeted cohorts, we curate 78 tasks for predicting the risk of a patient to progress from a disease to a more severe form. Our tasks span eight disease categories, and we predict time between diagnosis of the original disease and the progressed form (Figure 3e, Extended Data Table 8). The incidence for these tasks is higher than for the new onset tasks due to more restrictive inclusion criteria for the candidate set of censored patients (for more details see Downstream tasks in Online Methods). Similar to the new onset task formulation, we model the risk using a Cox-regression model with task performances computed over varying threshold durations, from 60 days (for acute diseases) up to five years (for more chronic diseases). Overall, A POLLO significantly outperforms the baseline on the large majority of tasks with 53 out of 78 diseases (p < 0.05). A POLLO shows the highest gains in risk prediction for disease progression for cardiovascular and circulatory diseases (on average from 0.65 to 0.75 AUROC), followed by the progression of diseases of the respiratory system and infectious diseases (on average from 0.63 to 0.75 AUROC), and the progression of cancers (on average from 0.62 to 0.74 AUROC). More specifically, A POLLO can predict the 5-year risk of hypertension to heart failure significantly better than the baseline (0.75 to 0.86), 3-year survival after stroke with 0.84 AUROC compared to 0.72 from the baseline, and 3-year survival after aortic stenosis with 0.83 compared to 0.72 for the baseline (all p < 0.0001). Moreover, A POLLO significantly improves the risk prediction for asthma exacerbation within 3 years (from 0.57 to 0.83 AUROC, p < 0.0001) and chronic obstructive pulmonary 9

disease (COPD) exacerbation within 3 years (from 0.57 to 0.79, p < 0.001). Finally, survival prediction significantly improves for melanoma from 0.71 to 0.87 (3-year survival) and for prostate cancer from 0.74 to 0.84 (5-year survival) (both p < 0.0001). It also improves for chronic myeloid leukemia from 0.75 to 0.83 (5-year survival), although this difference is not statistically significant due to relatively high variance across bootstraps. Results for all experiments can be found in Extended Data Table 9.

A POLLO predicts treatment response and risk of adverse events To evaluate how well A POLLO can model response to medications, we curate a set of 59 tasks for treatment response prediction spanning six disease categories, where the majority of treatments are neoplastic or immunomodulating agents (Figure 3f, Extended Data Table 10). Similar to the previous tasks, we use Cox regression to predict the risk for each endpoint starting from time of first treatment administration. Overall, A POLLO significantly outperforms the baseline on 30 out of 59 tasks (p < 0.05). Among the disease categories, the best performing groups are medications acting on the endocrine system, where A POLLO improves the risk prediction from 0.57 to 0.8 AUROC, and on the cardiovascular and circulatory system tract, where the performance increases from 0.64 to 0.75 AUROC using A POLLO embeddings (Figure 3f, Extended Data Table 10). In particular, A POLLO predicts stroke-free survival on apixaban therapy for atrial fibrillation with 0.81 AUROC significantly better than the baseline with 0.69 (p < 0.0001) and major adverse cardiac events (MACE)-free survival on losartan for hypertension with 0.8 AUROC (p < 0.0001) and on amlodipine for hypertension with 0.79 AUROC (p < 0.0001). For treatments of the respiratory system, A POLLO predicts the survival on tiotropium for COPD with 0.75 AUROC compared to 0.59 (p < 0.001) and exacerbation-free survival on umeclidinium/vilanterol for COPD with 0.71 AUROC compared to 0.6. Finally, for cancer treatments, A POLLO significantly improves survival prediction on trastuzumab and pertuzumab, both targeted therapies for HER2-positive breast cancers, from 0.66 to 0.93 (p < 0.0001) and from 0.75 to 0.89 (p < 0.05). Similarly, A POLLO significantly improves survival prediction for chemotherapy with paclitaxel for breast cancer (from 0.75 to 0.88, p < 0.01) and for hormone therapy with leuprolide for prostate cancer (from 0.65 to 0.86, p < 0.0001). Results for all experiments can be found in Extended Data Table 11. To assess the ability of A POLLO to anticipate clinically relevant adverse events, we evaluate a curated set of 17 tasks covering cardiovascular, gastrointestinal, metabolic, and drug-induced complications (Figure 3g, Extended Data Table 12). Across all adverse-event tasks, A POLLO significantly improves risk stratification on 12 out of 17 tasks over the baseline. For adverse events after nonsteroidal anti-inflammatory drug (NSAID) exposure, performance increases significantly for acute kidney injury (AUROC 0.80 to 0.91, p < 0.0001) and GI bleeding (0.80 to 0.91, p < 0.01). Similarly, the model captures metabolic risk with significant gains in predicting treatment-emergent type 2 diabetes following systemic corticosteroid use (0.70 to 0.87, p < 0.0001) and atypical antipsychotic therapy (0.69 to 0.84, p < 0.0001). Results for all experiments can be found in Extended Data Table 13. It is important to distinguish between the task of comparing across treatments versus making predictions across 10

patients in our scenario. Because each task conditions on a specific treatment, these predictions stratify within a treated cohort rather than across drugs. Therefore, the risk scores likely reflect a combination of baseline patient health and treatment-specific factors: patients at higher risk of adverse events from NSAIDs, for example, may have underlying conditions that predispose them to adverse events in general, rather than NSAID-specific risk.

A POLLO for clinical operations management We evaluate 12 operational tasks that reflect short-term in-hospital outcomes and resource demands (Figure 3h, Extended Data Table 14). These tasks include predicting all-cause inpatient readmission within 30 days, inpatient length of stay over 7 days, and 10 acute care endpoints where we predict whether the endpoint will occur within 7 days of emergency department (ED) admission. For the length of stay and acute care endpoints, we provide the patient history up to 24 hours after ED admission as input into the model, thus the effective threshold duration is 6 days. Since the censorship is substantial for these tasks, with incidences as low as 0.01%, we report performance using balanced accuracy (for more details see Section Metrics inOnline Methods). Overall, A POLLO yields significant improvements in 9 out of 12 settings. In particular, balanced accuracy increases from 0.65 to 0.97 for in-hospital dialysis dependence, 0.64 to 0.80 for acute kidney failure, and 0.66 to 0.79 for sepsis (all p < 0.0001). For broader operational metrics, the model improves prediction of prolonged hospital stay (0.63 to 0.76) and all-cause readmission within 30 days (0.54 to 0.62; both p < 0.0001). Results for all experiments can be found in Extended Data Table 15. Taken together, these findings indicate that the learned temporal patient representations generalize across pharmacologic settings and time scales, potentially providing a basis for the optimization of hospital operations in the future.

A POLLO’s risk scores stratify patient groups and are well-calibrated To assess whether the risk prediction with A POLLO can be used for patient stratification, we divide the patient cohort into low and high-risk groups based on the predicted risk scores. We select the top and bottom 25% of risk scores for the high- and low-risk groups, respectively (Extended Data Figures 1 to 8). Since the prevalence of many endpoints is very low in real-world scenarios, the stratification could be improved by using a fraction smaller than 25%. Across all sets of tasks, we observe that the risk scores of A POLLO stratify the patient cohorts well, particularly for tasks with lower censorship, e.g., cancer mortality, heart failure mortality (Extended Data Figure 3). Calibration metrics further confirm that the risk scores of the model are wellcalibrated across all groups of tasks, where we observe that tasks with larger sample sizes are better calibrated than tasks with fewer patients (Extended Data Figures 9 to 16).

Ablation studies Semantic alignment across modalities is crucial for modeling diseases with inherently multimodal diagnostics such as neoplasms. When assessing long term risk predictions on durations up to five years, understanding the patient health state across modalities and modeling the temporal context of the patient history is important. To 11

quantify the contribution of each component of our model, we perform a series of ablations on the neoplasm disease progression task set comprising 28 downstream tasks. We compare the full version of A POLLO to (i) a structured-data-only model trained with the same pretraining objectives, (ii) a supervised model with the same architecture trained end-to-end on each individual task, (iii) a progress-note baseline using the embedding of the most recent progress note, and (iv) the age-sex baseline (Figure 3i, Extended Data Figure 17). The strongest gains appear in hematologic malignancies: for chronic myeloid leukemia mortality, A POLLO improves over the progress note model by +0.31 AUROC and over the supervised model by +0.23. For solid tumors, A POLLO outperforms the supervised baseline by +0.16 on breast cancer palliative care, +0.13 on colorectal cancer mortality, and +0.12 on bladder cancer metastasis. Survival prediction for prostate cancer is already well captured by structured data alone (AUROC 0.83 vs. 0.84 with all modalities), whereas melanoma and breast cancer survival benefit more substantially from multimodal integration (+0.07 and +0.06 AUROC, respectively). These results show that the benefits of multimodal, temporally contextualized pretraining extend consistently across cancer types and clinical endpoints. Across all tasks, A POLLO achieves a mean AUROC of 0.735 improving over the structured-only variant by +0.025 (0.71). This highlights the importance of our multimodal integration incorporating clinical notes and imaging for modeling long-term cancer progression. The supervised baseline reaches a mean AUROC of 0.626 (-0.109), indicating that task-specific fine-tuning alone is insufficient compared to our pretraining strategy. Similarly, the progress note baseline obtains 0.615 on average (-0.12) and the age-sex baseline achieves 0.619 (-0.116), demonstrating that modeling temporal context beyond the most recent encounter is essential. Overall, the ablations confirm that multimodal integration, temporal modeling, and large-scale pretraining are all necessary components of A POLLO, each contributing substantially to accurate long-term cancer progression risk assessment.

Patient-level representations enable semantic search The prognostic evaluations above test whether A POLLO embeddings encode information predictive of future clinical events. A complementary question is whether the learned embedding space encodes similarities between patients based on past clinical events. Retrieval tasks probe this by querying the embedding space directly using nearest-neighbor search. We measure whether the returned patients have similar clinical profiles as the query. Strong retrieval performance would support the utility of A POLLO not only for prediction but also for cohort discovery, clinical trial matching, and medical reasoning. By compressing entire medical histories into a single vector, we can quantify the similarity of any two patients via cosine distance and thus perform semantic retrieval. To mimic a clinical deployment scenario, we construct a search index composed of the entire evaluation set of 1.4 million patients, each embedded at the calendar time of January 1st, 2025. We curated 61 patient cohorts defined by a particular diagnosis followed by administration of a first-line therapy. These inclusion criteria are extracted from the structured medical events via SQL. We divide each cohort into five equal folds, and set each member of one fold as the queries and retrieve the 12

ncology (

ancers)

Retrieval similarity

APO O

<ICD:Z94.0> (Kidney transplant status)

<Med:Tacrolimus> (immunosuppressive drug)

60-70 y.o. male: kidney transplant and history of

0.77

0.65

0.16

0.39

0.18

0.16

0.22

0.39

0.30

4 0.

0.30

0.19

0.16

End stage renal

disease

Kidney transplant status

hronic kidney disease, unspecified

N17.9

N18.9

M:Tac

N18.6

59.6

59.7

59.8

59.9

61.3

Hematology

Sensory

Z94

M:Tac

61.5

Dependence on renal dialysis

ther complication of kidney transplant

7k record

N18.6

M:Tac

4 4

T86.19

44 4

Z99 44

M:Tac

Z94.0&Tacrolimus

hronic kidney disease,

70+M, 53k record

60+M, 69k record

& Renal

M:Tac

Age

Hepatic

& Psychiatry

Neurology

astrointestinal,

& Autoimmune

Acute kidney failure, unspecified

Z94.0&Tacrolimus 60+M, 308k record

Tacrolimus

Musculoskeletal

Infectious

stage 3 unspecified

N18.30

N18.6

62.3

68.7

N17.9

0.880 60+y, male; ICD10: Atherosclerotic heart disease (I25),

Lab values: LDL 70+ mg/dL; Medications: Atorvastatin

LDL cholesterol ≥ 70 mg/dL

’ Pathology

60-70y, female

G b W O CG

w Z

; IDH

A-28-1751-01

73.3

65.

N18.6

73.3

73.7

51.5

66.6

Pathology G b V Array CGH S b w b ICD10 M b C 9 70+y, female;

M:Tac

73.3

0.826

grade I

ildtype

-00-D

T86.19

Encounter for palliative care

Z99 4

lastoma multiforme, grade I

Z94

68.7

Z99

Z99 4

N18.6

lio

hypertension (I10), pure hypercholesterolemia, unspecified (E78);

cardiovascular disease and

0.22

0.28

0.17

4 0.3

0.16

Respiratory

"History of atherosclerotic

44 0.

0.39

4 0.

olic

49 4

0.58

4 49 0.17

& Reproductive

50+M,

Last-progress-note

0.19

0.17

Endocrine, Meta

acute kidney failure and chronic kidney disease

0.56 0.32 0.19

0.37

0.31

0.20

ardiovascular irculatory

0.68 8

4 0.

0.38

0.28

4 0.2

0.31

0.28

0.5

0.77 7

4 0.

0.0

0.19

0.27

0.35

0.37

0.22

0.2

0.32

0.70

0.65

0.73

0.77 0.5

0.5

0.52

0.6

0.32

Acc@5

4 4

0.7

0.71

0.29

0.76

0.75

0.8

0.29

0.80

0.79

0.82

Apollo 0.81

0.85

1.0

0.82

despite statin therapy."

;

: "

lastoma,

ith glio

alignant neoplasm of

unspecified (

0.878

lio

71.

W O H

everal genomic

alances consistent

lastoma";

rain,

)

dL at age 25+"; Lab values: Total cholesterol: 200+mg/dL; LDL: 125+mg/dL; Triglycerides: 55+mg/dL; HDL: 65+mg/dL; Medications: Atorvastatin

0.876 50+y, male;

ICD10

: E78, I25;

: "

ossi

le familial

ositive family history of premature coronary heart disease (

achieving an LDL of 30-

APO O

Clinical Notes P

hypercholesterolemia. Untreated total cholesterol: 350+ mg/dL at age 25+.

Medications

0 mg/dL, I added

HD)" ; "

etia. Results very disappointing.";

ICD10

: E78;

Clinical Notes “P :

ossi

le familial

hypercholesterolemia, untreated LDL 180+ mg/dL. Elevated Lp(a).

C Medications on

restor 10 mg daily

” Lab values

ith good results ;

APO O LL

V GF

R amplification

loss of

0.811 60+y, male;

2A/2

0.808 70+y, male;

Image token at last time step

lastoma, :

ith E

, monosomy 10,

Pathology G b GM ICD10 C 9 :

lio

grade

lastoma,

W O H

o-copy

71.9

grade

T promoter

71.

Pathology G b GM

R-

olysomy 7,

RvIII detected, t

, IDH1 R132H negative, M

NGS MM P

: LDL: 100+ mg/dL;

lio

, IDH1 R132H negative; Arr

IDH1 R132H negative, M

urrently

: Rosuvastatin

Text token at last time step

Pathology G b W O ay CGH P w GF w C KN B ICD10 C

60+y, female;

unmethylated;

20+y, female;

0.815

: Rosuvastatin

0.875

Retrieval similarity

LLM

Retrieval similarity

30+y, male; ICD10: E78; Clinical Notes: "Untreated total cholesterol: 350+ mg/

lio

lastoma,

W O H

grade I

T promoter methylated;

roficient/MSS; Mutations in

TEN, TERT

promoter; gain on chromosome 7, including E

Figure 4: A POLLO enhances patient retrieval. (a) 61 patient retrieval tasks curated from combinations of ICD10 diagnosis codes and medications, assessed with accuracy among the five closest (Acc@5) embedded patients compared to retrieval of the latest progress note embedding, (b) qualitative evaluation of the closest embedded patient to a patient for kidney transplant maintenance. (c) Text-based retrieval on the example of a clinical trial inclusion criteria using A POLLO treatment assessment plan token as the last prompt. (d) Image-based retrieval using an external whole-slide image from TCGA using the corresponding TITAN embedding [43] as last token. To preserve patient privacy, ages shown in the figure are randomly shifted by -2 to 2 years from their actual age, and lab test and vitals values are presented as ranges instead of exact numbers.

most similar patients from the remainder of the search index (for more details see Section Retrieval in Online Methods). We compute the agreement between the top-5 semantically retrieved patients and the cohort defined by the structured inclusion criteria. As a baseline, we construct a search index using the unimodal embedding of the most recent progress note prior to 1/1/2025 (Figure 4a, Extended Data Figure 18). Compared to the baseline, embedding the entire patient history using A POLLO yields higher agreement with the SQL ground-truth cohorts across all tasks and achieves particularly high retrieval performance for diseases which more severely affect patient health. The top-performing cohorts include neoplasms (ovarian cancer on carboplatin, multiple myeloma on bortezomib, acute myeloid leukemia on cytarabine), circulatory diseases (chronic ischemic heart disease on aspirin, atrial fibrillation on apixaban, and hypertension on lisinopril), and metabolic and endocrine conditions (lipidemia on atorvastatin, hypothyroidism on levothyroxine, and T2DM on metformin). In contrast, lower retrieval performances are observed for less-serious conditions such as iron deficiency anemia or vitamin B12 deficiency anemia. These trends suggest that the patient state is primarily 13

impacted by conditions or interventions which have an impact on future diagnoses or mortality. Indeed, this is expected as the patient state is obtained from the masked-diagnosis prompt (Figure 1h). A full list of cohorts and their retrieval performance can be found in Extended Data Tables 16 and 17. We next take a deeper look at some qualitative retrieval results (Figure 4b). Our query patient is a 60-70 y.o. male with a history of chronic kidney disease and acute kidney failure, and was prescribed tacrolimus to prevent organ rejection following a kidney transplant (ICD10 code Z94.0). The top-4 closest patients (retrieved using A POLLO embeddings) have medical records ranging from 47 thousand to 308 thousand events. All four patients show histories of chronic kidney disease, acute kidney failure, and dependence on renal dialysis, which are very similar to the query patient. These results also show that semantic similarity in some cases can retrieve relevant patients which are missed by traditional SQL methods based on structured data. For example, the second patient is missing the ICD10 code for kidney transplant (Z94.0), despite having a history of kidney transplant complications (T86.19) and renal dialysis (Z99.2), followed by a prescription for tacrolimus. If we were to retrieve similar patients based only on the ICD10 code Z94.0, this patient would be missed. Besides retrieval using a query patient, A POLLO also enables retrieval using arbitrary modalities, in particular text (Figure 4c) and images (Figure 4d). For example, we can write a set of inclusion criteria in natural language and embed it as the query (Figure 4c). To generate the search index, we use the corresponding modality as the prompt token (i.e., masked text prompt instead of masked diagnosis prompt) to generate embeddings for every patient in the evaluation set. The top four retrieved patients all exhibit medical histories which align with the desired inclusion criteria, including high LDL cholesterol despite statin therapy, and likely or confirmed atherosclerotic heart disease. Additionally, we can use a single pathology slide as the query (Figure 4d). In this case, we use a slide from TCGA rather than from our evaluation set, to test the external generalizability of A POLLO embeddings. Similarly to the text-retrieval experiment, we utilize the masked image prompt token to generate the patient embeddings for the search index. The top retrieved patients all exhibit the pathology indicated in the TCGA slide: Grade IV Glioblastoma. Furthermore, at least three of the top 4 retrieved patients specifically align with the IDH wildtype status of the TCGA slide. The second patient is confirmed IDH wildtype due to absence of the R132H mutation and the presence of the +7/-10 chromosomal copy number variant signature. The third and fourth patients are IDH1 R132H negative which implies likely wildtype status. In summary, A POLLO patient embeddings unlock scalable semantic similarity search using diverse query modalities, ranging from a single textual description or image to an entire patient history.

A POLLO is interpretable at the bedside and at the population level A POLLO produces dynamic risk trajectories that track an individual’s probability of an outcome as new evidence accrues (Figure 5a–c). For three index patients, we plot 3-year risk of chronic kidney disease (CKD), lung cancer, and heart failure, against age. The curves show long periods of stability punctuated by stepwise increases that coincide with clinically meaningful events. To explain the increases, we perform a leave-onetoken-out (LOTO) sensitivity analysis over the interval that each jump happens. Specifically, we remove one event at a time from the interval and recompute the patient’s risk at the end of the interval. The largest posi14

Figure 5: A POLLO yields interpretable biomarkers at both the local and global level. (a–c) Local analysis. We plotted the model’s predicted 3-year risk for three example patients: (a) chronic kidney disease, (b) lung cancer, and (c) heart failure, as a function of age; markers indicate encounter times. At each prominent increase in risk, we performed a leave-one-token-out (LOTO) sensitivity analysis over events between the preceding visit t0 and the current visit t1 : tokens are removed one at a time and the change in predicted risk at t1 is recorded. Horizontal bars list the largest contributors in that interval, where positive values indicate increased risk and negative values decreased risk. (d–i) Global analysis. Population-level feature importance for six downstream tasks: (d) progression from COPD to respiratory failure, (e) heart failure to acute kidney failure, (f) type 2 diabetes to diabetic foot ulcer, (g) survival on trastuzumab for breast cancer, (h) atezolizumab for lung cancer, and (i) nivolumab for melanoma, obtained via Integrated Gradients (IG) as described in Online Methods. Bars show the top-10 tokens per task ranked by mean normalized IG among high-risk patients, and values are expressed as relative contribution (%). To preserve patient privacy, ages shown in the figure are randomly shifted by -2 to 2 years from their actual age, and lab test and vitals values are presented as ranges instead of exact numbers.

tive changes identify tokens that most contributed to the increase, and negative values identify tokens whose removal would have increased risk, indicating protective or competing signals. For onset of CKD (Figure 5a),

risk remained near zero through late 70s, then rose after documentation of essential hypertension, dyslipidaemia, and type 2 diabetes, and jumped again on a decline in eGFR to the low 50s and the findings of multiple cardiometabolic comorbidities, followed by repeated flank-pain evaluations. These tokens together reflect the emergence of chronic kidney damage on a background of long-standing vascular risk factors. For onset of lung cancer (Figure 5b), risk stayed low for decades, then increased with findings of smoking and prior breast cancer, and essential hypertension. Subsequent sharp rises came from COPD, emphysema and serial reports of an enlarging, FDG-avid lung nodule, mirroring the clinical work-up of incident lung cancer. For heart failure (Figure 5c), risk escalated following echocardiography evidence of severe aortic stenosis, ongoing smoking and COPD, and subsequent persistent refusal of valve intervention, progressive congestion and thrombosis. Additionally, the LOTO analysis also demonstrated protective events, such as borderline eGFR and normal serum alkaline phosphatase levels for CKD. It is worth noting that, besides canonical drivers, the top contributors also include medical events with clear associations with the diseases. While some of them are likely proxies such as markers of intensified surveillance or treatment, some reflect comorbidity clusters or shared risk environments, and a minority may represent hypothesis-generating signals for future study. We next quantify feature importance at the population level for high risk patients that fall in the top hazard quartile using Integrated Gradients (IG) computed on the pre-Transformer inputs and aggregated by net sum across embedding dimensions, normalization and max-pooled per patient, and averaged across the population (for more details see Section Interpretability in Online Methods). Across six representative tasks (Figure 5d-i), A POLLO recovers clinically coherent risk structure. For progression from COPD to respiratory failure (Figure 5d), high relative contributions are assigned to venous thromboembolism, tachycardia, abnormal lung imaging, severe hypertension and COVID-19, consistent with cardiopulmonary strain and intercurrent infection precipitating decompensation. Biomarkers including low serum iron and iron-deficiency anaemia are in line with recent works showing that iron deficiency in COPD is common and associated with worse exercise capacity, exacerbations, and prognosis [47, 48]. For progression from heart failure to acute kidney failure (Figure 5e), IG highlights obesity, post-procedural states and age proxies including seborrhoeic keratoses, nail disorders, osteoarthritis, vitreous body disease and zoster. While these likely partly reflect frailty and healthcare contact, several dermatologic and nail findings have been proposed as cutaneous markers of atherosclerotic and degenerative cardiovascular disease, including heart failure [49, 50]. For progression from type 2 diabetes to diabetic foot ulcer (Figure 5f), the dominant risk factors are atherosclerosis, vascular imaging, blood-pressure extremes and red-cell indices, reflecting macrovascular disease, haemodynamic load and systemic inflammation, all known to underlying diabetic foot pathology [51]. Gout also appears among the top contributors, which aligns with recent studies showing that gout in diabetes are linked to vascular complications, neuropathy and a high-risk diabetic foot, and that tophaceous gout can mimic or worsen foot ulcers [52, 53]. For survival on trastuzumab in breast cancer (Figure 5g), IG primarily surfaces host factors including high blood pressures, type 2 diabetes, chronic kidney disease and prior malignancy, consistent with data that cardiometabolic comorbidities and comorbidity burden influence both treatment tolerance and overall survival in HER2-positive breast cancer [54–56]. Symptoms such as mouth pain plausibly capture mucosal toxicity and treatment intensity over the course of therapy. For survival on atezolizumab in lung cancer and nivolumab in melanoma (Figure 5h,i), the dominant tokens include high blood pressures, diverticular disease, 16

kidney stones, pain and musculoskeletal diagnoses, as well as inflammatory skin conditions such as rosacea and hypertrophic skin disorders. Similarly, these patterns resonate with the facts that baseline comorbidity and functional status shape immunotherapy outcomes [55, 57], and with a growing literature linking cutaneous immune-related adverse events, including rosacea like eruptions and other inflammatory dermatoses to favourable survival on immune checkpoint inhibitors [58, 59]. To further investigate the generalization of these findings, we check additional local and global examples, including 3-year mortality in metastatic lung cancer and IG profiles for ovarian cancer metastasis, anthracycline cardiomyopathy, survival after aortic stenosis, and stroke/embolism-free survival with apixaban, all of which show similar concentration of attribution on clinically plausible drivers, multimorbidity clusters and care-process proxies (Extended Data Figure 19). Beyond these expected and recently reported associations, IG consistently surfaces tokens that are best interpreted as proxies for care processes such as post-procedural states and device-related codes, markers of multimorbidity and shared risk environments such as psychiatric, musculoskeletal or gastrointestinal diagnoses in cardiometabolic panels, and a small set of features with no obvious mechanistic link. We therefore treat these attributions as associative structure learned from the EHR rather than causal effects: they indicate which aspects of a patient’s record most drive predictions under A POLLO, highlight concordance with established risk factors and newer epidemiological observations, and nominate a handful of features such as iron deficiency in COPD or skin/nail aging markers in heart failure as testable, hypothesis-generating signals for future clinical studies. Together, the local and global analyses show that A POLLO’s predictions are auditable: at the bedside, clinicians can inspect why a patient’s risk rose when it did, and at scale, investigators can recover population-level structure that aligns with clinical knowledge while also revealing testable, hypothesis-generating signals. These properties are essential if foundation models are to support clinical decision-making and discovery rather than operate as opaque black boxes.

Discussion In this work, we introduce A POLLO, the first multimodal foundation model that integrates the complete longitudinal trajectory of patient data into a unified temporal patient representation. By training and evaluating on over 25 billion medical events spanning clinical notes, imaging, and structured records from more than 7 million patients, we demonstrate that it is possible to consolidate these fragmented data modalities from unimodal data silos into a single computationally meaningful representation space. This unified representation can form the basis of AI-enabled precision medicine, empowering a vast spectrum of clinical tasks from early disease detection to treatment response stratification. A POLLO’s virtual patient representations can continuously be updated with new clinical events, shifting healthcare from reactive, episodic treatment to proactive, continuous risk management. Moreover, these general-purpose representations also open new frontiers for clinical research. For example, by accurately identifying patients with similar complex multimodal phenotypes, A POLLO can readily enhance clinical trial matching, helping to resolve persistent bottlenecks in recruitment. For instance, APOLLO’s semantic retrieval identifies clinically similar patients even when structured coding is incomplete, as demonstrated by retrieval of a kidney transplant patient lacking the corresponding ICD-10 code (Figure 4b). In the longer term, this capability could lay the groundwork for next-generation trial designs, including in silico trial simulation at patient cohort level and highly personalized treatment response prediction at the individual level, potentially reducing the cost and duration of bringing new therapies to patients. Beyond predictive utility, A POLLO provides a glimpse of the potential for biomedical discovery from unified multimodal data. The model’s emergent embedding spaces, both at the input (medical event) and output (patient) level, reflects semantic relationships among individual medical concepts and patient-level phenotypes. The interpretability analyses, both at the patient level (via LOTO) and population level (via Integrated Gradients), provide some insight into model decision-making and lay the foundation for trustworthy application of multimodal risk-prediction models as an evolution of established clinical risk scores that use a limited set of manually curated features. Moving beyond manually curated feature sets is desirable not only because of the potentially higher accuracy as more input data is taken into account, but also because it provides a mechanism for identifying novel biomarkers and utilizing them for potential therapeutic benefit. Despite the potential of this technology, much work remains to be done. First, being trained on strictly observational data, A POLLO learns to make predictions that are associational, not causal. For example, treatment response prediction as evaluated here represents prognostic stratification across patients rather than efficacy estimation across different drugs. In other words, the model may primarily learn to predict which patients will respond to a particular drug, but not which drug will be the best choice for a given patient. This is due to the characteristics of our training data: the former capability is learned from patient timelines of multiple comparable patients for any given drug (which we have) while the latter requires outcome data of multiple comparable drugs for the same patient (which are far more difficult to obtain). While valuable for patient stratification, care must be taken not to interpret retrieved risk factors as direct targets for intervention without further causal validation. The counterfactual task of predicting differential outcomes across therapies for the same patient will be an important extension of this work. 18

On the data side, although the training dataset spans 17 hospitals and affiliated clinics, all data come from a single umbrella healthcare system (Mass General Brigham) operating on a shared Epic EHR platform. Accordingly, the present evaluation reflects within-system generalization. The patient population captured by MGB, while heterogeneous in many respects, is still primarily representative of the northeastern United States. Incorporating multi-institution data from additional health systems and care settings during training will be an important direction for broadening the geographic and operational scope of the learned representations. In the realm of risk prediction, the current benchmarking framework is designed to enable high-throughput evaluation across a broad and diverse set of 261 prognostic tasks, with age and sex serving as a common statistical reference. For well-studied endpoints such as heart failure, stroke, and acute MI, validated clinical risk tools (e.g., ASCVD, CHA2 DS2 -VASc, HEART score, Framingham) are the natural comparators. We do not perform head-to-head comparisons here, as computing these scores at scale requires endpoint-specific inputs that are not consistently captured as structured variables across the full cohort. However, our interpretability analyses recover many of the same risk factors embedded in these established scores (Figure 5), providing early evidence of mechanistic alignment and motivating direct comparisons in future work. On the modeling side, our design choices are necessarily constrained by computational feasibility when dealing with long sequences of comprehensive multimodal patient trajectory data. As a result, we use frozen, pre-trained, off-the-shelf unimodal encoders for computational efficiency. Future iterations that fine-tune or pre-train these encoders directly on in-domain EHR data would likely yield superior and more efficient representations. Similarly, we encode high-density modalities (such as long clinical notes) using basic aggregation techniques, such as averaging document chunk embeddings. Implementing more sophisticated, learnable aggregation rules could unlock more fine-grained signals. Finally, our study does not explicitly incorporate data streams from wearables, information about lifestyle, or patient-provider conversations. Future work should address these limitations via additional architectural innovations. Ultimately, the training paradigm and broad evaluation pipeline presented here demonstrate that converting the static EHR archive into unified, longitudinal virtual patient representations is feasible at healthcare-system scale, establishing a foundation for AI-enabled precision medicine across the full spectrum of clinical care.

Online Methods Dataset description We collected a large-scale, multimodal electronic health record (EHR) dataset, MGB-7M, from the Mass General Brigham (MGB) healthcare system. The dataset contains 7,155,044 patients with 25,296,943,893 clinical events in total across diverse modalities from 1992 to 2025, with the longest patient record spanning 33 years. MGB-7M is comprised of 28 distinct modalities, including structured records such as diagnosis, medications, lab tests, vital signs and flowsheet, as well as unstructured medical notes and imaging data. A detailed distribution of patient demographics and modalities of MGB-7M can be found in Extended Data Tables 1 and 2. The dataset was split into training, validation, and testing splits with a ratio of 75:5:20. The MGB institutional review board approved the retrospective analysis of the EHRs. The participants were not directly involved or recruited for the study. The requirement for informed consent to analyze the EHR data was thus waived.

Data acquisition and preprocessing Source systems and extraction. We extracted EHR data from the institutional analytics warehouse (Snowflake) using SQL queries against the operational Epic system. Raw tables were harmonized, de-identified according to institutional policy, and cleaned for format inconsistencies, duplicated rows, and implausible timestamps, before downstream processing. Modalities. The corpus comprised (i) demographic data including sex, ethnicity, and date of birth, (ii) temporally ordered structured data including diagnoses coded, medications coded, laboratory tests coded, vital signs, and flowsheet measurements, and (iii) temporally ordered unstructured data comprising clinical text and medical images, spanning multiple services as shown in Figure 1b-e. Imaging data included anatomic pathology whole slide images, gross images, electron-microscopy images, and hematology blood smear images, where we used both raw images and their diagnostic reports. For the remaining imaging services, only diagnostic reports were used. Preprocessing. For demographic data, we converted each patient’s record timestamp to minutes since birth and tokenized sex (male, female, unknown) as discrete categorical tokens. Ethnicity information in the EHR was recorded in a multi-select format where one patient could identify as multiple ethnicities. To handle this complication, we pre-embedded each atomic ethnicity value (e.g., “Ethiopian”) using bert-base-uncased [9] using the template “Ethnicity: {ethnicity}”. These contextual text embeddings allow us to take advantage of the world knowledge in existing pretrained general-purpose language models; in support of this hypothesis, when plotted via UMAP, the atomic ethnicity embeddings were loosely clustered by their associated geographic region. For patients who identified as multiple ethnicities, we took the mean of their atomic embeddings.

For structured data, we anchored each domain to standard clinical terminologies where possible: diagnoses were represented as ICD10 codes, medications were mapped to RxNorm ingredient identifiers, and vitals and laboratory tests were referenced using LOINC. Measurements with numeric values (laboratory tests, vital signs, and flowsheet entries) were first reconciled to a common unit per test, then transformed by sign(x) × log(|x| + 1) to stabilize heavy tails. To reduce the influence of outliers while retaining ordinal structure, we discretized each test into ten equal-width (post-transform) bins, where the central eight covered the empirical 2.5th –97.5th percentiles, and the remaining two captured extreme lower and upper values. Measurements with categorical values were canonicalized to a compact, test-specific answer set using an agentic workflow: one large language model proposed a canonical label inventory for each test and a second model mapped free-text responses (for example, “pos”, “positive”, “+”) to those labels. After normalization, we tokenized structured observations as follows: each ICD diagnosis and each RxNorm ingredient became a token; each numeric measurement was represented by a (code, quantile-bin) token; and each categorical measurement by a (code, canonical-category) token. This procedure yielded a structured vocabulary of 235,768 unique tokens with learnable embeddings similar as in large language models. Unstructured data were compressed to fixed-dimension embeddings using modality-specific pretrained encoders. All clinical text—including progress notes, diagnostic reports, and other note types were embedded with a pretrained clinical language model (Gatortron-base [18]). Notes exceeding the model’s context window were segmented into contiguous chunks and encoded independently, where chunk embeddings were then averaged to obtain a single note-level representation. For pathology and hematopathology, we retained both reports and images: whole-slide images were embedded with TITAN [43], hematology blood smear images with DinoBloom [60]; and electron microscopy and gross images with CONCHv1.5 [44]. For non-pathology imaging services, we used the accompanying diagnostic reports (text) rather than pixel data, providing a consistent representation across specialties.

Pretraining

Architecture. We modeled each patient as a single temporally ordered sequence of heterogeneous events, starting with their demographics. We embedded the patient’s sex (in {Male, Female, Unknown}) by a learned embedding layer W sex ∈ R3×E , where E denotes the model’s latent embedding dimension. The preprocessed ethnicity embedding was projected to the model’s latent space with a two-layer multilayer perceptron (MLP): Peth : RB → RE , where B = 768 denotes the embedding dimension of bert-base-uncased. Another twolayer MLP, Page : R+ → RE , was used for encoding patient’s age at last visit. The three demographics tokens, together with a learnable CLS token, were prepended to each patient’s timestamped records. For event t with timestamp τt (normalized to fraction of 100 years), we formed an input embedding zt ∈ RE by combining a content embedding with a learnable time encoding that replaces positional embeddings. Tokenized structured (k) events were embedded by a learned embedding layer W emb : N0 → RE . Let xt ∈ Rdk denote the preextracted embedding for unstructured event in modality k at time t, where dk is the embedding dimension of the modality-specific encoder for modality k. A lightweight projector Pk : Rdk → RE maps each unstructured 21

feature to the model’s latent space. A learnable time MLP ϕtime : R → RE encodes τt and is added to the content embedding. Thus,  Wtemb [ tokent ] + ϕtime (τt ) for structured modalities, zt = P x(kt ) + ϕ (τ ) for unstructured modalities, kt time t t

(1)

where W emb [tokent ] denotes indexing into the embedding matrix for the discrete token at time t. Each patient’s time-ordered sequence of events was then input into a multimodal temporal integration network, which is a Transformer encoder fθ [61]. We use the base configuration, with 12 transformer blocks, hidden layer dimension E=768, 12 attention heads of dimension 64, and a 4E-width MLP.

Masking and reconstruction objectives. Pretraining followed a multimodal masked modeling objective. For each patient, we randomly select a set of indices M i.i.d. with probability ρ = 0.3, and replaced their input with learned mask vectors. For structured tokens, reconstruction was formulated as classification within semantically coherent vocabularies. Masked structured tokens at index t ∈ M were replaced by an observation-type–specific mask vector mot ∈ RE , and decoding was performed within each modality. For laboratory, vital, and flowsheet measurements, we further constrained decoding by subdomains defined by LOINC CLASS. Laboratory measurements are assigned to classes using the official LOINC mapping; vital signs and flowsheet tests are assigned to analogous classes using a large language model, Qwen-32B [62], to harmonize local nomenclature. Decoder weights for discrete vocabularies were tied to the corresponding input embeddings. Let ht be the encoder state for a masked structured token and Vm the relevant (sub)modality vocabulary; then the pretraining objective is given by the cross-entropy loss Lstruct =

X 1 − log p(yt | ht ), |Mstruct | t∈M

p(· | ht ) ∈ ∆|Vmt |−1 .

(2)

struct

For unstructured events, reconstruction was treated as regression to the original encoder embedding. Masked (u) unstructured tokens were replaced by an input-type–specific mask vector mkt ∈ RE . The masked sequence Z̃ = {z̃t } was encoded to contextual states H = fθ (Z̃), with ht ∈ RE . A small, encoder-specific prediction head maps the contextual state ht back to the raw embedding space, and training minimizes a combination of mean-squared error and cosine distance: Lunstruct =

|Munstruct | t∈M

unstruct

1 (kt ) (kt ) 2 (kt ) (kt ) x̂ − xt + 1 − cos∠ x̂t , xt . 2 dkt t

The total objective is L = Lstruct + Lunstruct , computed as a global mean over masked tokens of each family.

(3)

Implementation details. We used a maximum sequence length of 1,536 events during pretraining. Sequences shorter than this were right-padded with an attention mask, and sequences longer than this were handled by uniformly sampling a contiguous window of 1,536 events per step. Optimization used AdamW with gradient clipping (max norm 1.0) and a cosine with warmup learning rate schedule. To stabilize optimization under multi-objective supervision and heterogeneous token availability, we employed two learning-rate groups: a lower rate for the shared encoder and embedding parameters, and a higher rate for the modalityspecific projectors and prediction heads. This choice mitigates high-variance updates from rare modalities while allowing small heads to adapt rapidly. Weight decay was applied to matrix-shaped parameters, and biases and LayerNorm weights were excluded. The model was trained for a maximum iterations of 30,000, with validating on the validation set every 1,000 iterations. The final model was selected based on the best validation loss. Training was done under distributed data parallelism on eight NVIDIA A100-80GB GPUs. Detailed hyperparameters used for model pretraining can be found in Extended Data Table 5.

Patient representation. Unless otherwise specified, we summarize a patient’s longitudinal record as a single vector by appending a diagnosis, i.e., ICD-10, mask token at inference time and assigning it the time encoding of the patient’s last visit. The augmented sequence is passed through the encoder, and the contextual state of this masked ICD token is taken as the patient embedding. Operationally, this “query” token asks the model to predict the patient’s current diagnostic profile, which is one of the objectives used during pretraining, thereby encouraging the embedding to aggregate those aspects of the history most predictive of present disease burden. This vector is then used as the patient representation for downstream analyses. For patients with extremely long histories, we capped the raw sequence to the most recent 100,000 events.

Evaluation framework We conducted time-to-event (TTE) evaluations across five primary categories: new disease onset, disease progression, treatment response, adverse event, and hospital operations. We evaluated A POLLO on 95 disease onset tasks (Extended Data Table 6), 78 disease progression tasks (Extended Data Table 8), 59 treatment response tasks (Extended Data Table 10), 17 adverse event tasks (Extended Data Table 12), and 12 hospital operations tasks (Extended Data Table 14). The metadata for each task can be found in the corresponding tables. In TTE tasks, the model (in our case, fixed A POLLO patient embeddings and a linear Cox head) receives as input all patient features up to a snapshot time, and aims to predict the duration between the snapshot time and a predetermined endpoint. For disease onset tasks, our objective was to predict the risk of a particular disease at the time of a hospital discharge (snapshot) prior to its diagnosis (endpoint). A higher risk corresponds to a shorter time to diagnosis, and a lower risk corresponds to a longer time to diagnosis. For disease progression tasks, we set the snapshot as the diagnosis of the disease of interest, and the endpoint is set to a proxy for disease progression. These proxies can include the diagnosis of a more severe disease variant, administration of a medication implying more severe disease, or death. For treatment response and adverse event tasks, the 23

snapshot is taken at time of first administration for a particular medication, and the endpoint is set to a proxy for treatment response, such as death or occurrence of an adverse event. We choose a TTE formulation for these tasks because they are subject to right-censorship, which occurs when a patient does not reach the endpoint as of the last event in their record. Censorship is influenced by both endpoint risk (i.e., low risk of reaching the endpoint means higher chance of censorship) as well as unrelated confounding factors (e.g., a patient moved to a different hospital). To evaluate model performance, we compute a variety of time-dependent discrimination and calibration metrics at a pre-specified threshold duration of τ ∈ R+ days (Extended Data Tables 7-17). The threshold duration is chosen individually for each task to reflect an interval in which an intervention applied at time 0 could plausibly change the outcome at timepoint τ (in days). For example, more chronic diseases such as heart disease may require a longer τ since interventions (e.g., lifestyle changes) may take a long time to manifest as appreciable changes in disease risk. On the other hand, more acute endpoints (e.g., mortality from COVID) require smaller τ since biological timescales are shorter. Practically, the choice of τ is also informed by the data distribution, such that sufficient cases are accumulated by time τ in order for metric calculations to be valid. For a detailed discussion of each metric, see Metrics & statistical analysis. When constructing the evaluation splits on the validation set of MGB-7M for each task, all valid snapshot and endpoint event times were extracted from each patient’s medical record. We then performed the following filtering steps to arrive at a single snapshot-endpoint pair for each patient, so that each patient is represented no more than once in each task split. For disease onset tasks, all hospital discharges after the first 5 visits are initially selected as valid snapshots and the first diagnosis of the disease was selected as the endpoint. To increase incidence of the endpoint and mimic how a real-world predictive model might be applied only to higher-risk patients, we kept only snapshots within two standard deviations of the mean age of diagnosis for the disease of interest. For disease progression tasks, the first diagnosis of the disease was selected as the snapshot and the first occurrence of the endpoint was selected. For treatment response tasks, the first administration of the treatment was selected as the snapshot. We kept the absolute first occurrence of the endpoint for all tasks unless otherwise specified below (see Task Definitions). We discarded the patient if no valid snapshot events existed prior to the endpoint event. We also discarded patients where the endpoint event occurred more than 100 years after the snapshot event. Furthermore, to filter out instances where snapshot and endpoint events occur close together due to potential administrative delays in documenting the snapshot event, we enforce a blackout period b where patients who experience the endpoint event within b days of the snapshot event are discarded. The duration of the blackout period depends on the time course of disease as defined by τ :   30 if τ > 90,   b=

7    1

if 60 ≤ τ < 90,

(4)

if τ < 60.

We further applied inclusion criteria for some disease onset tasks based on patient sex (see Task Definitions). After all filtering steps, if more than one valid snapshot-endpoint pair is present, a random one is selected for 24

each patient. We trained separate Cox proportional hazards models for each TTE task to estimate the risk of the endpoint event as a function of covariates. We extract patient-level features using A POLLO, where the input consists of all events in a patient’s medical record up to the snapshot time. For the structured modalities and last progress notes baselines (see Baselines), some patients may have zero valid events prior to the snapshot time; these patients are dropped to prevent them from adversely affecting baseline performance. For computational efficiency, the features are dimensionality-reduced using principal components analysis (PCA) to the top-50 principal components, which are used as the covariates x. The PCA projection is fit on the training set only (validation and test sets are transformed without refitting). The Cox model estimates a hazard function h(t|x) = h0 (t) exp(β ⊤ x),

(5)

where h0 (t) is a baseline hazard and β is the regression coefficients estimated from the training set. The resulting linear predictor β ⊤ x provides a continuous risk score used to rank patients by their relative risk of experiencing the endpoint. We use the CoxPHFitter implementation from lifelines 0.30.0, with the penalizer λ set to 10−4 as default and increased if necessary for convergence. Exact values of λ used for each experiment are specified in Extended Data Tables 7, 9, 11, 13, 15 and 17. Because many tasks have relatively rare endpoint events, we employed case-cohort sampling to improve computational efficiency and mitigate extreme censorship in the training data. Specifically, we constructed casecohort splits with a maximum ratio of 4:1 censored to noncensored individuals in the training set. This resampling alters the censorship ratio in the training set; however, the validation and test sets are left unmodified and retain the original, natural censorship ratio to ensure unbiased model evaluation. Because the Cox partial likelihood used for coefficient estimation is invariant to the baseline hazard, the modified training censorship does not bias the learned risk scores β ⊤ x. However, absolute survival probability estimates do depend on the baseline hazard, so this function was re-estimated on the unmodified validation set using the Breslow estimator prior to evaluation on the test set. This two-step procedure ensures that the regression coefficients benefit from the enriched case representation during training while the survival probability estimates reflect the true event distribution under the natural censoring regime.

Baselines

Age–sex baseline. To quantify the value of learned representations against a minimal set of demographics predictors, we constructed frozen patient features using only age and sex as covariates. Age was defined as the patient’s age (in days) at the time of the last visit included in the input sequence; sex was encoded as a 3-level categorical variable Male, Female, Unknown with dummy indicators. This baseline represents the information available in textbook risk scores that rely solely on demographics. During TTE evaluation, PCA was omitted for this baseline.

Structured modalities baseline. To better understand the contribution of unstructured modalities (notes and images) to a patient representation, we performed an ablation study by training a variant of A POLLO using only structured tokens (diagnoses, medications, labs, vitals, flowsheets) with the identical pretraining objective, tokenizer, and encoder as the full model. At inference, patient representations were computed from structured inputs only (using same ICD-10 mask-token prompt as for A POLLO). The resulting features are 768-dimensional, same as A POLLO.

Last progress notes baseline. To assess the benefit of temporal integration relative to a single text-only snapshot, we represented each patient by the embedding of their most recent progress note (by GatorTronbase [18]) prior to the prediction time. The resulting features are 1024-dimensional.

Supervised baseline. To evaluate the value of self-supervised pretraining, we trained the same transformer architecture end-to-end for each task without pretraining. Instead of the masked ICD query used by A POLLO to form a patient embedding, we used the CLS token and attached a two-layer MLP that outputs discrete hazards over time bins. Let hj ∈ (0, 1) denote the hazard for bin j (obtained by a sigmoid on the head logits), and Q Sj = k≤j (1 − hk ) the discrete survival through bin j. For an individual with event indicator δ ∈ {0, 1} and observed bin y, the negative log-likelihood is LNLL = −δ log Sy−1 + log hy − (1 − δ) log Sy ,

(6)

as in [63]. For each task with target horizon τ days, we defined the bin edges 0 = t0 < t1 < · · · < tJ < tJ+1 = ∞ as:

tj =

 {0} ∪ {365i : i = 1, . . . , 10} ∪ {∞}

ifτ > 365,

{0, 7, 15, 30, 90, 180, 365, ∞}

ifτ ≤ 365.

(7)

The head outputs one hazard per interior bin [tj , tj+1 ). We trained each model on a single GPU with batch size of 128, learning rate of 1e-5, optimized using AdamW optimizer. Each model was trained for a maximum 40 epochs, with early stopping on the validation set if no improvement on the validation loss was observed for 5 epochs.

Retrieval The retrieval tasks are set up as follows: a cohort is defined using a set of inclusion criteria and patients satisfying the inclusion criteria are extracted from the evaluation set of 1.4M patients using SQL queries.

Embeddings are extracted for all patients at a single calendar time (January 1st, 2025) to mimic the state of all patients on that day. Patients who are still alive as of that date have their medical records truncated such that only medical events prior to that date are included in the embedding. Patients who died prior to that date have all events in their medical history included in the embedding. For each task, the cohort is divided into five equal folds, and during each round of evaluation one fold (20% of the cohort) is removed from the search index to be used as the queries. Retrieval metrics are calculated on this fold, and we repeat the process with the remaining 4 folds to obtain mean and standard deviation performance. For the multimodal retrieval examples for text and image (Figure 4c,d and Extended Data 18) retrieval results, we retrieve from the entire evaluation set. The inclusion criteria for each cohort take the form of diagnosis THEN medication, where the medication is an approved first-line therapy for the diagnosis. The full list of cohorts and their descriptions is presented in Extended Data Table 16.

Interpretability To interrogate which inputs drive task-specific risk, we computed Integrated Gradients (IG) [64] from each patient’s pre-Transformer input sequence to the scalar output of the corresponding Cox head. Formally, letting s(·) denote the task head applied to the Transformer encoder fθ (·), and Z = {zt }Tt=1 ∈ RT ×E the input sequence immediately before the multimodal temporal modeling, we estimated token-level attributions Z 1 IG(Z) ≈ (Z − Z0 ) ⊙

∇Z s fθ Z0 + α, (Z − Z0 ) dα,

(8)

where Z0 is a baseline sequence of the same shape. The baselines were modality-consistent: for structured tokens we used the empirical mean of the learned embedding table (mean over the 235,768 unique structured tokens), and for unstructured inputs we used the all-zeros vector in the corresponding latent space. In present work, we primarily focus on risk factors, thus we restricted attribution to patients in the top quartile of predicted log-hazard for each task. This concentrates signal on clinically meaningful high-risk profiles and avoids diluting importances with low-risk trajectories. IG yields an E-dimensional attribution vector per token occurrence. We summarized each occurrence by the net P attribution (sum over embedding dimensions), at = E e=1 IGt,e . To mitigate scale differences across patients, , with ε > 0 a small constant to we normalized each patient’s token scores by their ℓ1 sum: Ãp,v = P AAp,v u p,u +ε avoid division by zero. For tokens that repeat within a patient, we applied occurrence-wise max pooling to obtain a single per-patient score per token: Ap,v = maxt∈T (p,v) at , where T (p, v) indexes all occurrences of token v for patient p. Population-level importance for token v was computed as the mean of per-patient normalized scores across

patients who had v at least once: Āv =

1 X Ãp,v . |Pv | p∈P

(9)

To suppress idiosyncratic or extremely rare features, we required v to meet a minimum prevalence in the analysis set: it must appear in ≥ 2.5% of either uncensored (event) or censored patients for that task. Tokens failing this criterion were excluded prior to ranking. For each task, tokens were ranked by Āv in descending order to produce a population-level list of riskincreasing features. Structured measurements are reported at the resolution used during modeling, i.e. (test, bin) for numeric data and (test, canonical category) for categorical entries, while diagnoses and medications are reported by their ICD-10 and RxNorm ingredient identifiers, respectively. Unstructured events were handled identically in the pipeline.

Metrics & statistical analysis For each metric, we report mean and 95% confidence interval computed across 100 non-parametric bootstraps of the test set. Below we define each of the metrics used. Cumulative/dynamic AUC. The cumulative/dynamic AUC quantifies the probability that, for a randomly selected pair of individuals (one who experiences the event before time τ and one who remains event-free beyond τ ) the predicted risk score is higher for the individual who fails earlier. To accommodate right-censored observations, this metric employs inverse probability of censoring weighting (IPCW), where the censoring survival function Ĝ(t) is estimated by the Kaplan-Meier curve of censoring times on the test set. Formally, this metric is represented by P P i j wij 1{r̂i > r̂j } P P , (10) AUC(τ ) = i j wij where wij are IPCW pair weights (estimated from Ĝ). We use the implementation from scikit-survival (sksurv.metrics.cumulative dynamic auc). Balanced accuracy. This is defined as the average of sensitivity and specificity for classifying whether a patient has reached the endpoint by τ . Patients censored prior to τ are removed. The binary risk threshold is chosen to maximize balanced accuracy on the validation set, then the metric is computed on the test set. Concordance index (c-index). We compute the truncated Uno’s c-index as implemented in sksurv.metri cs.concordance index ipcw, where τ is the truncation time. This metric provides a global summary of concordance over all pairs of events prior to τ . For two patients who both experience the event prior to τ , the c-index corresponds to the probability that the predicted risk score is higher for the patient who fails earlier. This metric uses the same censoring survival function Ĝ(t) as cumulative/dynamic AUC.

Brier score. Brier score quantifies the mean squared error (MSE) between predicted risk probabilities p̂i (τ ) and the observed event indicator at τ . To account for censoring we use the time-dependent IPCW Brier score implemented in sksurv.metrics.brier score: n 2 1X Brier(τ ) = wi (τ ) Yi (τ ) − p̂i (τ ) , n i=1

(11)

where Yi (τ ) = 1{Ti ≤ τ, δi = 1} is the event indicator by τ , and the IPCW weight is wi (τ ) =

1{Ti > τ } Ĝ(τ )

1{Ti ≤ τ, δi = 1} Ĝ(Ti− )

(12)

Here Ĝ(Ti− ) is the Kaplan–Meier estimate of the censoring survival function just before Ti , and Ti , δi are the observed time and event indicator for patient i. The IPCW weights correct the contribution of censored observations so that the sample mean approximates the expected squared error at τ . This metric uses the same censoring survival function Ĝ(t) as cumulative/dynamic AUC. Integrated Calibration Index (ICI). This is the mean absolute calibration error over the range of predicted probabilities. First we group the patients in the test set by predicted risk into 10 decile bins. For each bin k, we d (k) (τ ). compute the mean predicted risk probability p̂(k) and the Kaplan–Meier estimate of observed risk KM The ICI is given by K 1 X d KM(k) (τ ) − p̂(k) . (13) ICI = K k=1 Maximum Calibration Error (MCE). Similar to ICI, we compute MCE over the 10 bins of predicted risk. The maximum absolute calibration deviation across the range of predicted probabilities is given by d (k) (τ ) − p̂(k) . MCE = max KM k

(14)

Computing hardware and software We used Python (version 3.12.11) and PyTorch (version 2.6.0, CUDA 12.4) (https://pytorch.org/) for all experiments and analyses in the study, which can be replicated using open-source libraries as below. For model pretraining, we used 8 GPUs configured for multi-GPU training using distributed data-parallel (DDP). Our transformer additionally used flash-attn (version 2.7.4) and ninja (version 1.11.1.4). We used implementations, specifically PCA, StandardScaler and Logistic Regression from Scikit-learn (version 1.6.1) in our downstream tasks. We used the lifelines library (version 0.30.0) for its implementation of the Cox proportional hazards model in survival tasks. Matplotlib (version 3.10.3) and Seaborn (version 0.13.2) were used to create plots and figures. Use of other miscellaneous Python libraries is detailed in the Reporting 29

Summary.

Data availability Following institution policies, all requests for data collected or curated in-house will be evaluated on a case-bycase basis to determine whether the data requested is compliant with intellectual property and patient privacy obligations. Data can only be shared for academic research purposes and will require a material transfer agreement.

Code availability Code for the A POLLO pretraining and evaluation pipeline, including risk prediction and patient retrieval, will be made available for academic research purposes upon publication.

Author contributions A.Z., T.D., S.J.W., F.M. conceived the study and designed the experiments. A.Z., T.D., L.P.L., C.T., D.M., A.M., R.P. curated the dataset and performed data preprocessing. A.Z., T.D., M.Y.L. developed the A POLLO model architecture and pretraining pipeline. A.Z., S.J.W., C.T. performed the embedding space and patientlevel UMAP analyses. A.Z., T.D., S.J.W., C.T., J.E.L., R.P. curated and evaluated the prognostic downstream tasks. T.D., M.Y.L. performed the interpretability analysis. T.D., S.J.W. designed and conducted the multimodal retrieval experiments. A.Z., T.D., S.J.W., F.M. prepared the manuscript. All authors contributed to the writing. L.P.L., F.M. supervised the research.

Acknowledgments This work was funded in part by the Brigham and Women’s Hospital (BWH) and Mass General Hospital (MGH) internal funds.

References 1. Moor, M. et al. Foundation models for generalist medical artificial intelligence. Nature 616, 259–265 (2023). 2. Jensen, P. B., Jensen, L. J. & Brunak, S. Mining electronic health records: Towards better research applications and clinical care. Nature Reviews Genetics 13, 395–405 (2012). 3. RBC Capital Markets. The healthcare data explosion (2023). URL https://www.rbccm.com/en/ gib/healthcare/episode/the_healthcare_data_explosion. 4. Arcadia. Report: Only 57% of healthcare organizations’ data is used to make decisions. Tech. Rep., Healthcare Information and Management Systems Society (HIMSS) (2023). 5. Newman-Toker, D. E. et al. Burden of serious harms from diagnostic error in the USA. BMJ Quality & Safety 33, 109–120 (2024). 6. Cheng, Y., Wang, F., Zhang, P. & Hu, J. Risk prediction with electronic health records: A deep learning approach. In Proceedings of the 2016 SIAM international conference on data mining, 432–440 (SIAM, 2016). 7. Xiao, C., Choi, E. & Sun, J. Opportunities and challenges in developing deep learning models using electronic health records data: A systematic review. Journal of the American Medical Informatics Association 25, 1419–1428 (2018). 8. Brown, T. et al. Language models are few-shot learners. Advances in neural information processing systems 33, 1877–1901 (2020). 9. Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers), 4171–4186 (2019). 10. Oquab, M. et al. DINOv2: Learning robust visual features without supervision. Transactions on Machine Learning Research (2024). 11. Chen, T., Kornblith, S., Norouzi, M. & Hinton, G. A simple framework for contrastive learning of visual representations. In International conference on machine learning, 1597–1607 (PMLR, 2020). 12. Lin, Z. et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science 379, 1123–1130 (2023). 13. Nguyen, E. et al. Sequence modeling and design from molecular to genome scale with Evo. Science 386, eado9336 (2024). 14. Bommasani, R. et al. On the opportunities and risks of foundation models. ArXiv (2021). 15. Rajpurkar, P., Chen, E., Banerjee, O. & Topol, E. J. AI in health and medicine. Nature medicine 28, 31–38 (2022). 16. Tu, T. et al. Towards generalist biomedical AI. NEJM AI 1, AIoa2300138 (2024). 17. Alsentzer, E. et al. Publicly available clinical BERT embeddings. In Proceedings of the 2nd Clinical Natural Language Processing Workshop, 72–78 (Association for Computational Linguistics, Minneapolis, Minnesota, USA, 2019). URL https://www.aclweb.org/anthology/W19-1909. 18. Yang, X. et al. A large language model for electronic health records. npj Digital Medicine 5, 194 (2022). 31

19. Chen, R. J. et al. Towards a general-purpose foundation model for computational pathology. Nature Medicine 30, 850–862 (2024). 20. Vorontsov, E. et al. A foundation model for clinical-grade computational pathology and rare cancers detection. Nature Medicine 30, 2924–2935 (2024). 21. Pérez-Garcı́a, F. et al. Exploring scalable medical image encoders beyond text supervision. Nature Machine Intelligence 1–12 (2025). 22. Tiu, E. et al. Expert-level detection of pathologies from unannotated chest x-ray images via self-supervised learning. Nature Biomedical Engineering 6, 1399–1406 (2022). 23. Liu, S., Wang, X., Hou, Y. et al. Multimodal data matters: Language model pre-training over structured and unstructured electronic health records. IEEE Journal of Biomedical and Health Informatics 27, 504– 514 (2023). 24. Khader, F., Kather, J. N., Müller-Franzes, G. et al. Medical transformer for multimodal survival prediction in intensive care: Integration of imaging and non-imaging data. Scientific Reports 13, 10666 (2023). 25. Wornow, M., Thapa, R., Steinberg, E., Fries, J. & Shah, N. EHRshot: An EHR benchmark for few-shot evaluation of foundation models. Advances in Neural Information Processing Systems 36, 67125–67137 (2023). 26. Steinberg, E., Fries, J. A., Xu, Y. & Shah, N. MOTOR: A time-to-event foundation model for structured medical records. In ICLR (2024). 27. Li, Y. et al. BEHRT: Transformer for electronic health records. Scientific Reports 10, 7155 (2020). 28. Rasmy, L., Xiang, Y., Xie, Z., Tao, C. & Zhi, D. Med-BERT: Pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction. npj Digital Medicine 4, 86 (2021). 29. Miotto, R., Li, L., Kidd, B. A. & Dudley, J. T. Deep patient: An unsupervised representation to predict the future of patients from the electronic health records. Scientific Reports 6, 26094 (2016). 30. Redekop, E. et al. Zero-shot medical event prediction using a generative pretrained transformer on electronic health records. Journal of the American Medical Informatics Association 32, 1833–1842 (2025). 31. Renc, P. et al. Zero-shot health trajectory prediction using transformer. npj Digital Medicine 7, 256 (2024). 32. Zhang, Y. & Li, S. Chronoformer: Time-aware transformer architectures for structured clinical event modeling. arXiv preprint arXiv:2504.07373 (2025). 33. Kraljevic, Z., Bean, D., Shek, A. et al. Foresight—a generative pretrained transformer for modelling of patient timelines using electronic health records: A retrospective modelling study. The Lancet Digital Health 6, e281–e290 (2024). 34. Shmatko, A., Jung, A. W., Gaurav, K. et al. Learning the natural history of human disease with generative transformers. Nature (2025). 35. Li, Y., Mamouei, M., Salimi-Khorshidi, G. et al. Hi-BEHRT: Hierarchical transformer-based model for accurate prediction of clinical events using multimodal longitudinal electronic health records. IEEE Journal of Biomedical and Health Informatics 27, 1106–1117 (2023). 36. Kauffman, J. et al. Embedding methods for electronic health record research. Annual Review of Biomedical Data Science 8 (2025). 37. Soenksen, L. R. et al. Integrated multimodal artificial intelligence framework for healthcare applications. npj Digital Medicine 5, 149 (2022). 32

38. Huang, Z., Bianchi, F., Yuksekgonul, M., Montine, T. J. & Zou, J. A visual-language foundation model for pathology image analysis using medical Twitter. Nature Medicine 29, 2307–2316 (2023). 39. Adduri, A. K. et al. Predicting cellular responses to perturbation across diverse contexts with STATE. bioRxiv 2025.06.26.661135 (2025). 40. Bunne, C. et al. How to build the virtual cell with artificial intelligence: Priorities and opportunities. Cell 187, 7045–7063 (2024). 41. Johnson, A. E. W. et al. MIMIC-IV, a freely accessible electronic health record dataset. Scientific Data 10, 1 (2023). 42. Organization, W. H. ICD-10: International statistical classification of diseases and related health problems: Tenth revision (2004). 43. Ding, T. et al. A multimodal whole-slide foundation model for pathology. Nature Medicine 1–13 (2025). 44. Lu, M. Y. et al. A visual-language foundation model for computational pathology. Nature Medicine 30, 863–874 (2024). Publisher: Nature Publishing Group. 45. Pellegrini, C., Özsoy, E., Bani-Harouni, D., Keicher, M. & Navab, N. From ehrs to patient pathways: Scalable modeling of longitudinal health trajectories with llms. arXiv preprint arXiv:2506.04831 (2025). 46. Waxler, S. et al. Generative medical event models improve with scale. arXiv preprint arXiv:2508.12104 (2025). 47. Amado, C. A. et al. Non-anemic iron deficiency predicts COPD exacerbations and hospitalizations: Results from a prospective cohort. Journal of Clinical Medicine 14, 4154 (2025). 48. Dong, Z. et al. Association between iron homeostasis and prognosis in patients with chronic obstructive pulmonary disease: A retrospective analysis from MIMIC-IV database. Frontiers in Medicine 12, 1610681 (2025). 49. Ghonemy, S., Nasr, M. M. M., Soliman, M. & Hosiney, H. A. Clinical skin aging score and risk of degenerative cardiovascular diseases. The Journal of Clinical and Aesthetic Dermatology 14, 34 (2021). 50. Katira, A. & Katira, R. Dermatological manifestations of cardiac conditions. The British Journal of Cardiology 29, 9 (2022). 51. Soyoye, D. O., Abiodun, O. O., Ikem, R. T., Kolawole, B. A. & Akintomide, A. O. Diabetes and peripheral artery disease: A review. World Journal of Diabetes 12, 827 (2021). 52. Rasmussen, C., Larsen, J. W., Holm, P. C. & Nielsen, G. L. Gout: An overlooked disease in patients with diabetes? a danish prospective cohort study with 2 years of follow-up. Clinical Diabetes 43, 282–290 (2025). 53. Rasmussen, C. et al. Identifying tophaceous gout in foot ulcers using ulcer debris microscopy in type 2 diabetes. Journal of Wound Management 26, 175–181 (2025). 54. Valiyaveettil, D., Joseph, D. & Malik, M. Cardiotoxicity in breast cancer treatment: Causes and mitigation. Cancer Treatment and Research Communications 37, 100760 (2023). 55. Yalçıner, M. et al. Impact of comorbidity on survival in cancer patients receiving immune checkpoint inhibitors. Clinical and Translational Oncology 1–8 (2025). 56. Carreira, H. et al. Use of anthracyclines and trastuzumab for breast cancer in women with and without a history of cardiovascular disease in sweden: A national cross-sectional study. Cardio-Oncology 11, 56 (2025). 33

57. Poletto, S. et al. Predictive factors in metastatic melanoma treated with immune checkpoint inhibitors: From clinical practice to future perspective. Cancers 16, 101 (2023). 58. Du, Y., Wu, W., Chen, M., Dong, Z. & Wang, F. Cutaneous adverse events and cancer survival prognosis with immune checkpoint inhibitor treatment: A systematic review and meta-analysis. JAMA Dermatology 159, 1093–1101 (2023). 59. Cho, Y.-T., Lin, Y.-T., Yang, C.-W. & Chu, C.-Y. Cutaneous immune-related adverse events among taiwanese cancer patients receiving immune checkpoint inhibitors link to a survival benefit. Scientific Reports 12, 7021 (2022). 60. Koch, V. et al. DinoBloom: A foundation model for generalizable cell embeddings in hematology. In International Conference on Medical Image Computing and Computer-Assisted Intervention, 520–530 (Springer, 2024). 61. Vaswani, A. et al. Attention is all you need. Advances in neural information processing systems 30 (2017). 62. Yang, A. et al. Qwen3 technical report. arXiv preprint arXiv:2505.09388 (2025). 63. Zadeh, S. G. & Schmid, M. Bias in cross-entropy-based training of deep survival networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 43, 3126–3137 (2020). 64. Sundararajan, M., Taly, A. & Yan, Q. Axiomatic attribution for deep networks. In Precup, D. & Teh, Y. W. (eds.) Proceedings of the 34th International Conference on Machine Learning, vol. 70 of Proceedings of Machine Learning Research, 3319–3328 (PMLR, 2017). 65. Chase, D. M., Mahajan, A., Scott, D. A., Hawkins, N. & Kalilani, L. The impact of varying levels of residual disease following cytoreductive surgery on survival outcomes in patients with ovarian cancer: A meta-analysis. BMC Women’s Health 24, 179 (2024). 66. Petrucelli, N., Daly, M. B. & Pal, T. BRCA1- and BRCA2-associated hereditary breast and ovarian cancer. In Adam, M. P., Bick, S., Mirzaa, G. M. et al. (eds.) GeneReviews® [Internet] (University of Washington, Seattle, Seattle (WA), 2025). Initial posting: 1998-09-04. Updated: 2025-03-20. PMID: 20301425. Bookshelf ID: NBK1247. 67. Praestegaard, C. et al. Cigarette smoking is associated with adverse survival among women with ovarian cancer: Results from a pooled analysis of 19 studies. International Journal of Cancer 140, 2422–2435 (2017). 68. Neuendorff, N. R. et al. Anthracycline-related cardiotoxicity in older patients with acute myeloid leukemia: A young SIOG review paper. Blood Advances 4, 762–775 (2020). 69. Wang, J. et al. Impact of chronic kidney disease on the prognosis of transcatheter aortic valve replacement in patients with aortic stenosis: A meta-analysis of 133624 patients. Annals of Thoracic and Cardiovascular Surgery 28, 83–95 (2022). 70. Wallis, C. J. et al. Association between use of antithrombotic medication and hematuria-related complications. JAMA 318, 1260–1271 (2017).

Acute Lymphocytic Nephrotic Syndrome Leukemia (n=320,401, Aplastic Anemia Addison's Disease Cushing Syndrome Ankylosing Spondylitis (n=324,298, I=0.0%) I=0.0%) (n=319,560, I=0.0%) (n=316,111, I=0.0%) (n=300,909, I=0.0%) (n=299,891, I=0.0%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00000 1.0000 1.0000 1.0000 1.000 1.000 0.99975

0.9995

0.99950

0.9995

0.9990

0.999

0.9990

0.999 0.998

0 121 243 365 0 608 12161825 0 365 730 1095 0 608 12161825 0 365 730 1095 0 608 12161825 Amyotrophic Lateral Chronic Myeloid Leukemia Myasthenia Gravis Systemic Sclerosis Sclerosis (n=237,830, Oropharyngeal Cancer Pyelonephritis (n=293,012, I=0.0%) (n=285,573, I=0.0%) (n=275,207, I=0.0%) I=0.0%) (n=206,439, I=0.0%) (n=109,969, I=0.1%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.001 1.0000 1.000 1.000 1.000 1.000 1.0000

Disease-free probability Disease-free probability

Disease-free probability

Extended Data Figures

0.9995

0.999

0 608 12161825 0 365 730 1095 Immune Thrombocytopenia Epilepsy (n=107,104, (n=109,648, I=0.1%) I=0.4%) p < 0.0001 p < 0.0001 1.000 1.000 0.995

0.999 0 121 243 365

0.999

0.998

0.98

0.998

0.9995

0.9990 0 608 12161825 0 608 12161825 0 608 12161825 0 121 243 365 Pneumonia (n=106,143, Iron Deficiency Anemia Glioblastoma (n=104,795, Celiac Disease I=2.2%) (n=106,088, I=2.3%) I=0.1%) (n=104,737, I=0.2%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.00 1.000 1.000

0.990

0.998

0.996

Ulcerative Colitis (n=103,581, I=0.2%) p < 0.0001 1.000

0 121 243 365 0 121 243 365 0 121 243 365 0 365 730 1095 Inflammatory Bowel Disease (n=103,506, Crohn's Disease Acute Pancreatitis Hyperthyroidism I=0.4%) (n=102,575, I=0.2%) (n=102,428, I=0.4%) (n=101,992, I=0.4%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.000 1.0000 1.000

0.995

0.99

Disease-free probability Disease-free probability

Disease-free probability

0.9990

0.999

0.995

0 365 730 1095 Generalized Anxiety Disorder (n=100,308, I=2.3%) p < 0.0001 1.00

0.9975

0.995

0.99

0.9950

0.990

0.98

0 365 730 1095 0 365 730 1095 0 365 730 1095 0 121 243 365 0 365 730 1095 0 121 243 365 Systemic Lupus Psoriasis (n=100,293, Erythematosus (n=99,078, Schizophrenia (n=98,823, Intracranial Hemorrhage Asthma (n=96,555, Thyroid Cancer I=0.5%) I=0.1%) I=0.1%) (n=97,919, I=0.4%) I=2.8%) (n=96,533, I=0.2%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.000 1.000 1.0000 1.00 1.000 0.99

0.998

0.995

0.999

0.9975

0.998

0.9950

0.95

0.995

0.990 0 365 730 1095 0 608 12161825 0 121 243 365 0 121 243 365 0 365 730 1095 0 608 12161825 All Leukemia (n=95,941, Bipolar Disorder Hepatitis B (n=95,382, Sleep Apnea (n=95,353, Acute Myeloid Leukemia Sensorineural Hearing I=0.2%) (n=95,481, I=0.5%) I=0.1%) I=3.7%) (n=94,240, I=0.1%) Loss (n=93,646, I=1.3%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.000 1.000 1.000 1.00 1.000 1.00

Disease-free probability Disease-free probability

0.990

0.95 0.998

0.990

0.90

0.998

0.98

0.96 0 608 12161825 0 365 730 1095 0 365 730 1095 0 365 730 1095 Non-Alcoholic Fatty Vitamin B12 Deficiency Depression (n=93,151, Hepatitis C (n=93,125, Liver Disease (n=92,929, Hypothyroidism Anemia (n=92,631, Rheumatoid Arthritis I=5.2%) I=0.2%) I=1.5%) (n=92,851, I=2.3%) I=0.1%) (n=92,124, I=0.4%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.000 1.00 1.00 1.000 1.00 0 608 12161825

0.95

0 121 243 365

0.995 0.990

0.98

0.95

0.96

0 121 243 365 0 608 12161825 0 365 730 1095 Venous Thromboembolism Retinal Detachment Multiple Sclerosis (n=92,077, I=2.0%) (n=91,313, I=0.1%) (n=91,280, I=0.1%) p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.000 1.0000 0.98

0.999

0.9975

0.96

0.998

0.9950

0 121 243 365 Days after snapshot

0 365 730 1095 0 121 243 365 Chronic Otitis Media Migraine (n=90,133, (n=90,662, I=1.0%) I=2.3%) p < 0.0001 p < 0.0001 1.00 1.00 0.99

0 608 12161825 Days after snapshot

0.999

0 121 243 365 Days after snapshot

0.98 0 121 243 365 Days after snapshot

0.99 0 365 730 1095 Melanoma (n=88,489, I=0.4%) p < 0.0001 1.00 0.99 0 608 12161825 Days after snapshot

Extended Data Figure 1: Kaplan-Meier curves for new disease onset tasks using the top 25% (red) and bottom 25% (blue) predicted risk scores. Tasks are sorted by number of events (n) in the test set per task. Statistical significance is given by the p-value for every task; n: number of events; I: label imbalance; τ : task duration.

Disease-free probability

0.999 0.998

0.9975

0.995

0.9950

0.990

0 365 730 1095

0 608 1216 1825

0.98

0.99

0.995

Colorectal Cancer (n=82,952, I=0.3%) p < 0.0001 1.000 0.995

0.99

0.990

Disease-free probability

0.98 0 121 243 365

0.9975

0.95

0 365 730 1095

0.98 0 608 1216 1825

Glaucoma (n=79,639, Hidradenitis Suppurativa All-Cause Mortality I=2.0%) (n=79,200, I=0.2%) (n=78,240, I=7.7%) p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.0000 1.0 0.9

0 121 243 365 Interstitial Lung Disease (n=76,277, I=0.7%) p < 0.0001 1.00 0.98

0.9950 0 608 1216 1825 0 365 730 1095 0 121 243 365 0 608 1216 1825 Acute Myocardial Infarction (n=74,297, Non-Melanoma Skin Cancer Multiple Myeloma Aortic Aneurysm I=1.1%) (n=73,798, I=1.9%) (n=71,663, I=0.2%) (n=70,080, I=1.0%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.000 1.000 1.00 0.975

0.99

0.995

0.950

0.98

0.98 0.96

0.995

0.9975

0.95

0.9950 0 608 1216 1825

Kidney Cancer (n=80,008, I=0.3%) p < 0.0001 1.00

0 608 1216 1825

Type 2 Diabetes (n=76,085, I=3.4%) p < 0.0001 1.0

Liver Cancer (n=74,739, I=0.2%) p < 0.0001 1.000 0.995

0.9

0.990 0 365 730 1095

0 608 1216 1825

Chronic Kidney Disease Pancreatic Cancer (n=69,163, I=4.7%) (n=69,000, I=0.3%) p < 0.0001 p < 0.0001 1.0 1.0000 0.9

0.975

0.98

0.9

0.950

0.96

0.8

0 608 1216 1825

0 365 730 1095

0.99

0 121 243 365 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 0 365 730 1095 Chronic Obstructive Chronic Lymphocytic Pulmonary Disease Leukemia (n=68,605, Aortic Stenosis Lung Cancer (n=67,433, Osteoarthritis (n=68,687, I=1.5%) I=0.1%) (n=67,486, I=1.1%) I=0.9%) (n=67,258, I=6.6%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.0000 1.000 1.00 1.0

0.9950 0 243 486 730 Dyslipidemia (n=66,743, I=13.7%) p < 0.0001 1.0 0.8

Osteoporosis (n=66,705, I=3.5%) p < 0.0001 1.0 0.9

0.9

0 608 1216 1825 Parkinson's Disease (n=61,132, I=0.5%) p < 0.0001 1.00

0.8 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 0 608 1216 1825 0 608 1216 1825 Hypertension (n=61,086, Macular Degeneration Ovarian Cancer Breast Cancer (n=48,093, Alzheimer's Disease I=10.7%) (n=60,845, I=2.5%) (n=50,889, I=0.2%) I=2.2%) (n=44,219, I=0.8%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.00 1.000 1.00 1.00

0.95

0.8

0.98

0.90 0 608 1216 1825

0 365 730 1095

Endometriosis (n=43,788, Endometrial Cancer I=0.8%) (n=41,462, I=0.4%) p < 0.0001 p < 0.0001 1.00 1.00

0 608 1216 1825 Polycystic Ovary Syndrome (n=28,485, I=1.1%) p < 0.0001 1.00

0.99

0.98

0.99

0.98 0 365 730 1095 Days after snapshot

0 608 1216 1825 Days after snapshot

0 730 1460 2190 0 608 1216 1825 Chronic Ischemic Heart Disease (n=65,113, Atrial Fibrillation I=4.9%) (n=64,564, I=5.0%) p < 0.0001 p < 0.0001 1.0 1.0

0.9975

0 608 1216 1825 Peripheral Artery Heart Failure (n=66,617, Disease (n=65,448, I=3.5%) I=1.4%) p < 0.0001 p < 0.0001 1.0 1.00

Disease-free probability Disease-free probability

Disease-free probability

0 608 1216 1825

0.99

Disease-free probability

0.98

0 121 243 365

0.96

0 608 1216 1825 Primary Cirrhosis (n=82,336, Stroke (n=81,579, Hyperparathyroidism Gout (n=82,453, I=1.1%) I=0.5%) I=1.7%) (n=80,549, I=0.2%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.00 1.00 1.000

Disease-free probability

Meniere Disease Sarcoidosis (n=87,733, Hip Fracture (n=86,501, Peripheral Neuropathy Non-Hodgkin Lymphoma Hyperprolactinemia (n=88,001, I=0.1%) I=0.1%) I=0.7%) (n=85,988, I=1.3%) (n=85,695, I=0.3%) (n=83,097, I=0.2%) p < 0.01 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.000 1.0000 1.000 1.00 1.00 1.000

0.9

Bladder Cancer (n=64,368, I=0.4%) p < 0.0001 1.00 0.99

0.8

0.995

0.95

0.990

0 608 1216 1825 0 608 1216 1825 Benign Prostatic Hyperplasia (n=25,344, Prostate Cancer I=6.7%) (n=24,948, I=3.1%) p < 0.0001 p < 0.0001 1.0 1.00 0.9

0.98 0.96 0 608 1216 1825

0.95 0.90

0.96 0 365 730 1095 Days after snapshot

0.9

0 365 730 1095

0 365 730 1095 Days after snapshot

0 608 1216 1825 Days after snapshot

Extended Data Figure 2: Kaplan-Meier curves for new disease onset tasks. Continued. 36

0.975

0.98

0.950

0.96 0

608 1216 1825

0.99 608 1216 1825

0.995

0.998 0

121 243 365

0.990

0.99 0.98

608 1216 1825

0.9

0.95 0

365 730 1095

608 1216 1825

Type 2 Diabetes to Type 2 Diabetes to Diabetic Nephropathy Diabetic Neuropathy (n=21,252, I=1.6%) (n=20,438, I=5.0%) p < 0.0001 p < 0.0001 1.00 1.000 0.975 0.99 0.950 0.98

0.98 0

Asthma to Exacerbation Requiring Oral Corticosteroids (n=24,162, I=7.8%) p < 0.0001 1.00

608 1216 1825

0.995 0

Obesity to Type 2 Diabetes (n=24,164, I=9.1%) p < 0.0001 1.0

608 1216 1825

0 608 1216 1825 Chronic Ischemic Heart Episodic to Chronic Disease to Acute Migraine (n=13,503, NAFLD to Liver Fibrosis NAFLD to NASH (n=12,761, Myocardial Infarction I=0.8%) (n=12,813, I=0.4%) I=0.7%) (n=12,530, I=4.5%) p < 0.0001 p < 0.001 p = 0.126 p < 0.0001 1.000 1.000 1.000 1.00

0.996

GERD to Barrett's Esophagus (n=29,798, I=1.5%) p < 0.0001 1.00

0.96

365 730 1095

Type 2 Diabetes to Insulin Use (n=19,528, I=23.0%) p < 0.0001 1.0 0.9

608 1216 1825

365 730 1095

Chronic Ischemic Heart Atrial Fibrillation to Disease to Heart Failure Ischemic Stroke (n=12,096, I=12.6%) (n=10,916, I=3.6%) p < 0.0001 p < 0.0001 1.0 1.00 0.98 0.9 0.96

0 365 730 1095 0 365 730 1095 0 365 730 1095 Benign Prostatic Osteoporosis to Bone Inflammatory Bowel Hyperplasia to Urinary COVID-19 Mortality Fracture (n=10,084, Glaucoma to Blindness Disease to Colorectal Acute COVID-19 to Long Retention (n=7,929, (n=10,165, I=0.5%) I=1.2%) (n=9,408, I=1.1%) Cancer (n=8,794, I=0.5%) COVID (n=8,165, I=0.7%) I=4.9%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.01 p < 0.001 p < 0.0001 1.000 1.00 1.00 1.000 1.00 1.00 0.98 0.998 0.995 0.99 0.99 0.99 0.96 0.996 0.990 0 20 40 60 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 0 121 243 365 0 365 730 1095 Prediabetes to Type 2 Heart Failure to Acute Diabetes (n=7,776, Heart Failure Mortality Breast Cancer Mortality Breast Cancer Palliative Breast Cancer Metastasis Kidney Failure (n=6,354, I=6.2%) (n=6,806, I=29.8%) (n=6,736, I=12.7%) Care (n=6,733, I=2.9%) (n=6,514, I=4.4%) I=9.6%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.0 1.0 1.000 1.00 1.00 0.975 0.8 0.95 0.95 0.9 0.95 0.950 0.6 0.90 0.90 0.8 0 365 730 1095 0 365 730 1095 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 0 121 243 365 COPD to Chronic COPD to Pulmonary Chronic Hepatitis or Respiratory Failure Hypertension (n=5,788, Fatty Liver to Cirrhosis CKD Stage 3 to Stage 4+ COPD to Exacerbation Stroke Mortality (n=5,858, I=2.8%) I=3.0%) (n=5,786, I=3.4%) (n=5,442, I=4.6%) (n=5,425, I=5.5%) (n=5,239, I=18.8%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.00 1.00 1.00 1.00 1.0 0.98 0.98 0.95 0.8 0.95 0.95 0.96 0.96 0.90 0 365 730 1095 0 365 730 1095 0 608 1216 1825 0 365 730 1095 0 365 730 1095 0 365 730 1095 Barrett's Esophagus to Prostate Cancer Prostate Cancer Thyroid Cancer Psoriasis to Psoriatic Esophageal Cancer Mortality (n=4,533, Metastasis (n=4,426, Paroxysmal to Persistent Metastasis (n=3,701, Arthritis (n=3,521, (n=5,018, I=1.0%) I=16.2%) I=3.5%) Afib (n=4,176, I=2.8%) I=2.7%) I=4.5%) p < 0.01 p < 0.0001 p < 0.0001 p < 0.001 p < 0.0001 p < 0.01 1.00 1.0 1.00 1.00 1.000 1.000 0.975 0.975 0.8 0.99 0.95 0.98 0.950 0.950 0.6 0.98 0 608 1216 1825 0 608 1216 1825 0 608 1216 1825 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 Non-Proliferative to Rheumatoid Arthritis to Bladder Cancer Proliferative Diabetic Interstitial Lung Rheumatoid Arthritis to Metastasis (n=3,491, Retinopathy (n=2,921, Lung Cancer Mortality Disease (n=2,686, Second-Line Therapy Lung Cancer Metastasis I=3.9%) I=3.7%) (n=2,701, I=39.2%) I=2.3%) (n=2,681, I=6.5%) (n=2,534, I=9.0%) p < 0.0001 p < 0.001 p < 0.0001 p = 0.492 p < 0.0001 p < 0.0001 1.00 1.000 1.00 1.00 1.00 1.0

Disease-free probability

Disease-free probability Disease-free probability Disease-free probability Disease-free probability

Disease-free probability

Hypertension to Chronic Hypertension to Heart Hypertension to Atrial Kidney Disease Stage 3+ Failure (n=59,782, Fibrillation (n=57,973, (n=60,041, I=6.9%) I=6.1%) I=8.1%) p < 0.0001 p < 0.0001 p < 0.0001 1.00 1.00 1.00 0.95 0.95 0.95 0.90 0.90 0.90 0 608 1216 1825 0 608 1216 1825 0 608 1216 1825 Diabetes Mellitus to Peripheral Artery Diabetes Mellitus to Type 2 Diabetes to Disease (n=22,170, Diabetic Retinopathy Diabetic Foot Ulcer I=3.3%) (n=21,954, I=4.4%) (n=21,465, I=0.9%) p < 0.0001 p < 0.0001 p < 0.0001 1.000 1.00 1.00

0.95 0.90

0 365 730 1095 Days after snapshot

0.975

0.75

0.950

0.50 0 365 730 1095 Days after snapshot

0 243 486 730 Days after snapshot

0.9

0.95

0.98 0 608 1216 1825 Days after snapshot

0.90

0 365 730 1095 Days after snapshot

0.8

0 243 486 730 Days after snapshot

Extended Data Figure 3: Kaplan-Meier curves for disease progression tasks using the top 25% (red) and bottom 25% (blue) predicted risk scores. Tasks are sorted by number of events (n) in the test set per task. Statistical significance is given by the p-value for every task; n: number of events; I: label imbalance; τ : task 37 duration.

Disease-free probability Disease-free probability

1.0

Pancreatic Cancer Metastasis (n=2,169, I=4.8%) p < 0.0001

1.0

Colorectal Cancer Mortality (n=2,053, I=23.9%) p < 0.0001

1.0

0.8

0.9 0

365 730 1095

0.8

0.6

0.95

243 486 730

Melanoma Mortality (n=1,974, I=13.4%) p < 0.0001

0.95

0.75

0.90

0.9

365 730 1095

1.0

Ovarian Cancer Metastasis (n=1,967, I=4.9%) p < 0.0001

0.9

0.9 0

1.0

Colorectal Cancer Metastasis (n=1,913, I=6.5%) p < 0.0001

0 365 730 1095 0 365 730 1095 Pulmonary Fibrosis to Melanoma Metastasis Hepatitis C to Cirrhosis All Leukemia Mortality Liver Cancer Metastasis Pulmonary Hypertension Cirrhosis to Liver (n=1,905, I=5.2%) (n=1,601, I=6.6%) (n=1,451, I=25.2%) (n=1,404, I=3.6%) (n=1,369, I=6.2%) Cancer (n=1,293, I=3.7%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.01 p < 0.01 p = 0.420 1.00 1.00 1.00 1.00 1.00 1.00

365 730 1095

0.95

0.50

0.90

0.9

0.8

0.98

0.96 0 243 486 730 0 608 1216 1825 0 365 730 1095 0 243 486 730 0 365 730 1095 0 365 730 1095 Cirrhosis to Esophageal Myelodysplastic Syndrome Gestational Diabetes to Varices (n=1,252, to Leukemia (n=1,248, Thyroid Cancer Mortality Bladder Cancer Mortality Type 2 Diabetes Kidney Cancer Mortality I=6.9%) I=8.5%) (n=1,246, I=7.9%) (n=1,187, I=25.2%) (n=1,150, I=7.2%) (n=869, I=20.8%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p = 0.085 p < 0.0001 1.0 1.0 1.0 1.0 1.0 1.0

0.90

Disease-free probability Disease-free probability Disease-free probability

1.0

Aortic Stenosis Mortality (n=2,449, I=23.4%) p < 0.0001

0.9 0.8

1.00

365 730 1095

NASH to Cirrhosis (n=852, I=6.5%) p < 0.001

0.95 0.90

0.8

243 486 730

365 730 1095

608 1216 1825

365 730 1095

Pancreatic Cancer Kidney Cancer Metastasis CKD Stage 2 to Stage 3 Mortality (n=766, (n=826, I=8.7%) (n=794, I=17.1%) I=50.8%) p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.0 1.0 0.9

0.8 0 243 486 730 Chronic Lymphocytic Leukemia Mortality (n=618, I=18.0%) p < 0.0001

0.5

0.8

0.9

0.6

1.00 0.95

608 1216 1825 Mild Cognitive Impairment to Alzheimer's Disease (n=698, I=6.9%) p = 0.050

0.6

365 730 1095

Mild Cognitive Impairment to Dementia (n=674, I=12.0%) p < 0.0001 1.0 0.9

0.90

0.8 0 608 1216 1825 0 243 486 730 0 365 730 1095 0 365 730 1095 Chronic Myeloid Leukemia Acute Lymphocytic Acute Myeloid Leukemia Mortality (n=492, Liver Cancer Mortality Leukemia Mortality Mortality (n=343, I=16.7%) (n=472, I=40.3%) (n=412, I=21.8%) I=42.9%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.00 1.0 1.0

Ovarian Cancer Mortality (n=670, I=31.9%) p < 0.0001 1.00 1.0 0.75 0.8 0.75 0.8 0.8 0.5 0.50 0.6 0.50 0.6 0.6 0 365 730 1095 0 608 1216 1825 0 608 1216 1825 0 121 243 365 0 243 486 730 0 243 486 730 Days after snapshot Days after snapshot Days after snapshot Days after snapshot Days after snapshot Days after snapshot

Extended Data Figure 4: Kaplan-Meier curves for disease progression tasks. Continued.

Disease-free probability Disease-free probability Disease-free probability Disease-free probability Disease-free probability Disease-free probability Disease-free probability Disease-free probability

MACE-Free Survival on MACE-Free Survival on MACE-Free Survival on Amlodipine for Lisinopril for Aspirin for Chronic Survival on Furosemide Hypertension (n=5,059, Hypertension (n=4,986, Ischemic Heart Disease for Heart Failure I=21.5%) I=20.1%) (n=3,491, I=26.8%) (n=2,779, I=34.7%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.0 1.0 1.0 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0 365 730 1095 0 365 730 1095 0 365 730 1095 0 243 486 730 Survival on Metoprolol Stroke/Embolism-Free ESRD-Free Survival on MACE-Free Survival on Succinate for Heart Survival on Apixaban for Lisinopril for Chronic Clopidogrel for Chronic Failure (n=2,427, Atrial Fibrillation Kidney Disease (n=1,028, Ischemic Heart Disease I=34.4%) (n=2,007, I=23.4%) I=27.2%) (n=1,006, I=27.0%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.00 1.0 1.0 0.8 0.8 0.8 0.75 0.6 0.6 0.6 0.50 0

MACE-Free Survival on MACE-Free Survival on Losartan for Atorvastatin for Chronic Hypertension (n=2,738, Ischemic Heart Disease I=18.3%) (n=2,558, I=28.1%) p < 0.0001 p < 0.0001 1.0 1.00 0.75 0.8 0.50 0.6 0 365 730 1095 0 365 730 1095 HF-Free Survival on Survival on Lisinopril Empagliflozin for Type 2 for Heart Failure Diabetes (n=972, (n=981, I=31.8%) I=14.2%) p < 0.0001 p < 0.0001 1.0 1.0 0.8 0.8 0.6

243 486 730

0 365 730 1095 0 365 730 1095 0 365 730 1095 0 243 486 730 0 121 243 365 Stroke/Embolism-Free Stroke/Embolism-Free MACE-Free Survival on Survival on Trastuzumab Survival on Warfarin for Survival on Survival on Leuprolide Survival on Rivaroxaban Ticagrelor for Chronic for Breast Cancer Atrial Fibrillation Spironolactone for Heart for Prostate Cancer for Atrial Fibrillation Ischemic Heart Disease (n=923, I=13.2%) (n=895, I=38.4%) Failure (n=893, I=30.1%) (n=768, I=20.7%) (n=568, I=25.0%) (n=544, I=18.6%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.00 1.00 1.0 1.0 1.0 0.8 0.75 0.8 0.8 0.75 0.8 0.6 0.50 0.6 0.6 0.50 0.6 0 365 730 1095 0 365 730 1095 0 243 486 730 0 243 486 730 0 365 730 1095 0 365 730 1095 Exacerbation-Free Survival on Tiotropium Survival on Cisplatin Survival on Paclitaxel Survival on Carboplatin Survival on Pertuzumab Survival on Rituximab for COPD (n=542, for Head & Neck Cancer for Breast Cancer for Lung Cancer (n=446, for Breast Cancer for Non-Hodgkin Lymphoma I=40.4%) (n=499, I=17.4%) (n=490, I=17.3%) I=56.3%) (n=432, I=8.3%) (n=408, I=23.0%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.0 1.00 1.00 1.0 1.00 0.75 0.75 0.75 0.8 0.5 0.8 0.50 0.50 0.50 0 365 730 1095 0 243 486 730 0 365 730 1095 0 121 243 365 0 365 730 1095 0 365 730 1095 MACE-Free Survival on Survival on Fluorouracil Chlorthalidone for Survival on Carboplatin Survival on Sacubitril / Survival on Nivolumab Survival on for Colorectal Cancer Hypertension (n=344, for Endometrial Cancer Valsartan for Heart for Melanoma (n=304, Pembrolizumab for Lung (n=360, I=41.4%) I=17.7%) (n=338, I=30.2%) Failure (n=324, I=17.3%) I=29.3%) Cancer (n=252, I=57.1%) p < 0.0001 p < 0.001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.0 1.00 1.00 1.0 1.0 0.8 0.8 0.75 0.75 0.5 0.5 0.6 0.6 0.50 0.50 0 243 486 730 0 365 730 1095 0 243 486 730 0 243 486 730 0 243 486 730 0 121 243 365 Survival on Lactulose Survival on Exacerbation-Free Survival on Rituximab Survival on for Hepatic Pembrolizumab for Survival on Carboplatin Survival on Umeclidinium for Diffuse Large B-Cell Pembrolizumab for Breast Encephalopathy (n=243, Melanoma (n=235, for Ovarian Cancer / Vilanterol for COPD Lymphoma (n=214, Cancer (n=247, I=21.1%) I=30.9%) I=28.9%) (n=225, I=42.2%) (n=222, I=31.1%) I=25.2%) p < 0.0001 p < 0.01 p < 0.001 p < 0.001 p < 0.0001 p < 0.0001 1.0 1.00 1.0 1.0 1.0 1.00 0.8 0.8 0.75 0.75 0.5 0.5 0.6 0.6 0.50 0.50 0 365 730 1095 0 243 486 730 0 243 486 730 0 243 486 730 0 365 730 1095 0 243 486 730 Survival on Cytarabine for Acute Myeloid Survival on Paclitaxel Survival on Oxaliplatin Survival on Gemcitabine Survival on Cisplatin Survival on Carboplatin Leukemia (n=210, for Ovarian Cancer for Stomach Cancer for Pancreatic Cancer for Bladder Cancer for Esophageal Cancer I=41.4%) (n=207, I=38.2%) (n=196, I=52.0%) (n=191, I=68.1%) (n=188, I=26.6%) (n=187, I=41.2%) p < 0.01 p < 0.0001 p < 0.0001 p = 0.060 p < 0.001 p < 0.01 1.0 1.0 1.0 1.0 1.00 1.00 0.75 0.75 0.5 0.5 0.5 0.5 0.50 0.50 0.0 0 243 486 730 0 243 486 730 0 243 486 730 0 121 243 365 0 243 486 730 0 243 486 730 Survival on Sildenafil for Pulmonary Arterial Survival on Doxorubicin Survival on Bortezomib Survival on Paclitaxel Survival on Atezolizumab Survival on Octreotide Hypertension (n=187, for Soft Tissue Sarcoma for Multiple Myeloma for Pancreatic Cancer for Lung Cancer (n=159, for Neuroendocrine I=37.4%) (n=181, I=55.2%) (n=171, I=31.6%) (n=161, I=70.8%) I=62.3%) Tumors (n=148, I=31.1%) p < 0.0001 p < 0.0001 p < 0.01 p = 0.097 p = 0.021 p < 0.01 1.0 1.0 1.0 1.0 1.0 1.00 0.8 0.75 0.5 0.5 0.5 0.5 0.6 0.50 0 365 730 1095 0 243 486 730 0 243 486 730 0 121 243 365 0 121 243 365 0 365 730 1095 Days after snapshot Days after snapshot Days after snapshot Days after snapshot Days after snapshot Days after snapshot

Extended Data Figure 5: Kaplan-Meier curves for treatment response tasks using the top 25% (red) and bottom 25% (blue) predicted risk scores. Tasks are sorted by number of events (n) in the test set per task. Statistical significance is given by the p-value for every task; n: number of events; I: label imbalance; τ : task 39 duration.

Disease-free probability Disease-free probability

Survival on Azacitidine for Acute Myeloid Leukemia (n=111, I=57.7%) p = 0.117

Survival on Riluzole for Survival on Abiraterone Survival on Temozolomide Survival on Azacitidine Amyotrophic Lateral for Prostate Cancer for Glioblastoma (n=100, for Myelodysplastic Sclerosis (n=95, (n=100, I=53.0%) I=71.0%) Syndrome (n=99, I=56.6%) I=45.3%) p = 0.013 p = 0.093 p = 0.692 p = 0.021 1.0 1.0 1.0 1.0 1.0 0.75 0.5 0.5 0.50 0.5 0.5 0.5 0.25 0 243 486 730 0 243 486 730 0 243 486 730 0 121 243 365 0 243 486 730 0 243 486 730 Survival on Survival on Ipilimumab Pembrolizumab for Survival on Osimertinib Survival on Pemetrexed Survival on Enfortumab for Kidney Cancer (n=84, Colorectal Cancer (n=81, for Lung Cancer (n=79, for Mesothelioma (n=75, vedotin for Bladder I=41.7%) I=43.2%) I=55.7%) I=60.0%) Cancer (n=72, I=36.1%) p = 0.016 p = 0.143 p = 0.021 p = 0.061 p = 0.413 1.0 1.0 1.0 1.0 1 Survival on Nivolumab for Kidney Cancer (n=140, I=45.0%) p < 0.01

0.5

0.5 0 243 486 730 Days after snapshot

0 243 486 730 Days after snapshot

0.5 0 121 243 365 Days after snapshot

0 243 486 730 Days after snapshot

Heparin-Induced Acute Kidney Injury Tardive Dyskinesia on Type 2 Diabetes After Rhabdomyolysis on Thrombocytopenia GI Bleeding on NSAIDs After NSAIDs (n=65,679, Antipsychotics Systemic Corticosteroids Statins (n=58,987, (n=112,549, I=0.1%) (n=66,042, I=0.5%) I=1.8%) (n=64,815, I=0.1%) (n=64,078, I=2.2%) I=0.2%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0000 1.000 1.000 1.000 1.00 1.000 0.995

0.999

0.9998

0.999

0.98

0.998

0.96

0.999

0.998 0.990 0 4 9 14 0 30 60 90 0 10 20 30 0 243 486 730 0 243 486 730 0 121 243 365 Angioedema on ACE GI Bleeding on Clostridioides difficile Hyperkalemia on ACE Major Bleeding on Oral Type 2 Diabetes After Inhibitors (n=29,245, Antiplatelet Agents Infection on PPIs Inhibitors (n=9,647, Anticoagulants (n=8,387, Atypical Antipsychotics I=0.3%) (n=26,571, I=1.7%) (n=22,134, I=0.8%) I=2.4%) I=4.4%) (n=7,738, I=2.4%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.000 1.00 1.000 1.000 1.00 1.00 0.995 0.98 0.99 0.995 0.95 0.999 0.990 0.96 0.98 0.990

Disease-free probability

Extended Data Figure 6: Kaplan-Meier curves for treatment response tasks. Continued.

0 30 60 90 Hyperthyroidism on Amiodarone (n=7,323, I=0.7%) p < 0.01 1.00

0 121 243 365 0 121 243 365 0 10 20 30 0 121 243 365 Cardiomyopathy After Hypoglycemia on Hypothyroidism on MACE on Levothyroxine Anthracyclines (n=3,190, Sulfonylureas (n=2,918, Amiodarone (n=2,193, with History of Heart I=3.4%) I=2.4%) I=3.9%) Failure (n=602, I=38.5%) p < 0.0001 p < 0.0001 p < 0.001 p < 0.0001 1.0 1.00 1.00 1.00

0.99

0.9 0 243 486 730 Days after snapshot

0.98 0 608 1216 1825 Days after snapshot

0 121 243 365 Days after snapshot

0.95

0.75

0.90

0.50 0 243 486 730 Days after snapshot

243 486 730

0 243 486 730 Days after snapshot

Extended Data Figure 7: Kaplan-Meier curves for adverse events tasks using the top 25% (red) and bottom 25% (blue) predicted risk scores. Tasks are sorted by number of events (n) in the test set per task. Statistical significance is given by the p-value for every task; n: number of events; I: label imbalance; τ : task duration.

Disease-free probability

Acute Respiratory Acute Clostridioides Acute Venous Acute Liver Failure Distress Syndrome Acute Cardiac Arrest difficile Infection Thromboembolism Acute Sepsis (n=59,932, (n=64,401, I=0.1%) (n=219,603, I=0.0%) (n=70,065, I=0.2%) (n=67,118, I=0.4%) (n=61,349, I=2.4%) I=1.4%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 +1 1.0000 1.00000 1.000 0.0000 1.000 1.000 0.999 0.99975 0.0002 0.998 0.9998 0.995 0.998 0.99950 0.0004 0.996 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 Acute Dialysis Acute CatheterInpatient Readmission Dependence (n=53,315, Acute Pressure Ulcer Acute Kidney Failure Inpatient Length of Stay Associated UTI (All Cause) (n=13,718, I=0.4%) (n=51,579, I=0.7%) (n=51,353, I=4.4%) (n=45,508, I=100.0%) (n=44,623, I=0.1%) I=100.0%) p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.0 1.0 1.000 1.000 1.00 1.0000

0 2 4 6 Days after snapshot

0.5

0.99

0.998

0.999

0 2 4 6 Days after snapshot

0.98

0 2 4 6 Days after snapshot

0.0

0.8

0.9995

0.6 0 2 4 6 Days after snapshot

0 2 4 6 Days after snapshot

0 10 20 30 Days after snapshot

Extended Data Figure 8: Kaplan-Meier curves for clinical operation management tasks using the top 25% (red) and bottom 25% (blue) predicted risk scores. Tasks are sorted by number of events (n) in the test set per task. Statistical significance is given by the p-value for every task; n: number of events; I: label imbalance; τ : task duration.

Nephrotic Syndrome (n=324,298, I=0.0%, 365d)

ICI: 0.000

Addison's Disease (n=316,111, I=0.0%, 5y) ICI: 0.000

Cushing Syndrome (n=300,909, I=0.0%, 3y) ICI: 0.000

Observed risk

ICI: 0.000

Aplastic Anemia (n=319,560, I=0.0%, 3y)

0.04 MCE: 0.001

0.04 MCE: 0.002

0.04 MCE: 0.001

0.02

Observed risk

ICI: 0.000

Acute Lymphocytic Leukemia (n=320,401, I=0.0%, 5y)

0.04

0.00 0.00 0.00 0.02 0.04 0.00 0.02 0.04 Chronic Myeloid Leukemia Myasthenia Gravis (n=293,012, I=0.0%, (n=285,573, I=0.0%, 5y) 3y) ICI: 0.000 MCE: 0.004

0.02

Ankylosing Spondylitis (n=299,891, I=0.0%, 5y) 0.06 ICI: 0.001 0.04

MCE: 0.006

0.02

0.00 0.00 0.00 0.00 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.000 0.025 0.050 Systemic Sclerosis Amyotrophic Lateral Oropharyngeal Cancer Pyelonephritis (n=275,207, I=0.0%, Sclerosis (n=237,830, (n=206,439, I=0.0%, (n=109,969, I=0.1%, 5y) I=0.0%, 5y) 5y) 365d)

ICI: 0.000

0.04 MCE: 0.001

0.04

0.02

ICI: 0.000 MCE: 0.001

0.04

ICI: 0.000 MCE: 0.001

0.02

0.04

ICI: 0.000 MCE: 0.001

0.04

ICI: 0.001 MCE: 0.006

0.02

ICI: 0.003

0.075 MCE: 0.015 0.050 0.025 0.000 0.00 0.05

Observed risk

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 Immune Thrombocytopenia Iron Deficiency Anemia Celiac Disease (n=109,648, I=0.1%, Epilepsy (n=107,104, Pneumonia (n=106,143, (n=106,088, I=2.3%, Glioblastoma (n=104,795, (n=104,737, I=0.2%, 365d) I=0.4%, 365d) I=2.2%, 365d) 365d) I=0.1%, 3y) 3y) ICI: 0.000 ICI: 0.001 ICI: 0.002 ICI: 0.002 ICI: 0.001 ICI: 0.002 0.10 0.06 0.06 0.075 MCE: 0.005 MCE: 0.008 MCE: 0.006 MCE: 0.011 0.10 MCE: 0.011 0.04 MCE: 0.002 0.04 0.04 0.050 0.05 0.05 0.02 0.02 0.02 0.025 0.00 0.000 0.00 0.00 0.00 0.00 0.00 0.02 0.04 0.00 0.05 0.00 0.05 0.10 0.00 0.05 0.10 0.000 0.025 0.050 0.000 0.025 0.050 Ulcerative Colitis Inflammatory Bowel Crohn's Disease Acute Pancreatitis Hyperthyroidism Generalized Anxiety (n=103,581, I=0.2%, Disease (n=103,506, (n=102,575, I=0.2%, (n=102,428, I=0.4%, (n=101,992, I=0.4%, Disorder (n=100,308, 3y) I=0.4%, 3y) 3y) 365d) 3y) I=2.3%, 365d) 0.10 ICI: 0.002 0.075 ICI: 0.001 0.06 ICI: 0.001 ICI: 0.002 ICI: 0.002 0.075 ICI: 0.001 0.075 MCE: 0.014 MCE: 0.007 MCE: 0.002 MCE: 0.009 MCE: 0.005 0.06 MCE: 0.008 0.050 0.04 0.050 0.050 0.04 0.05 0.025 0.02 0.025 0.025 0.02 0.00 0.00 0.000 0.00 0.000 0.000 0.00 0.05 0.00 0.05 0.10 0.00 0.05 0.000 0.025 0.050 0.00 0.05 0.00 0.05 Systemic Lupus Intracranial Hemorrhage Psoriasis (n=100,293, Erythematosus (n=99,078, Schizophrenia (n=98,823, (n=97,919, I=0.4%, Asthma (n=96,555, Thyroid Cancer I=0.5%, 3y) I=0.1%, 5y) I=0.1%, 365d) 365d) I=2.8%, 3y) (n=96,533, I=0.2%, 5y) 0.075 ICI: 0.002 0.001 ICI: 0.000 0.001 ICI: 0.004 0.002 0.06 ICI: 0.06 ICI: 0.075 ICI: MCE: 0.005 MCE: 0.011 MCE: 0.002 MCE: 0.008 MCE: 0.010 0.15 MCE: 0.011 0.04 0.050 0.04 0.04 0.050 0.10 0.02 0.025 0.02 0.02 0.025 0.05 0.000 0.00 0.00 0.00 0.00 0.000 0.00 0.05 0.000 0.025 0.050 0.00 0.02 0.04 0.000 0.025 0.050 0.0 0.1 0.00 0.05 Bipolar Disorder Sensorineural Hearing All Leukemia (n=95,941, (n=95,481, I=0.5%, Hepatitis B (n=95,382, Sleep Apnea (n=95,353, Acute Myeloid Leukemia Loss (n=93,646, I=1.3%, I=0.2%, 5y) 365d) I=0.1%, 5y) I=3.7%, 3y) (n=94,240, I=0.1%, 3y) 3y) ICI: 0.002

0.075 MCE: 0.015 0.050 0.025 0.000 0.00 0.05

0.04 0.02

ICI: 0.001 MCE: 0.005

ICI: 0.004

0.2 MCE: 0.021 0.1

0.04 0.02

ICI: 0.001 MCE: 0.003

ICI: 0.002

0.10 MCE: 0.013 0.05

0.00 0.0 0.00 0.00 0.00 0.02 0.04 0.0 0.1 0.2 0.00 0.02 0.04 0.00 0.05 0.10 Non-Alcoholic Fatty Vitamin B12 Deficiency Depression (n=93,151, Hepatitis C (n=93,125, Liver Disease (n=92,929, Hypothyroidism Anemia (n=92,631, Rheumatoid Arthritis I=5.2%, 365d) I=0.2%, 5y) I=1.5%, 3y) (n=92,851, I=2.3%, 3y) I=0.1%, 365d) (n=92,124, I=0.4%, 3y) 0.2 0.06 0.15 ICI: 0.003 ICI: 0.001 ICI: 0.002 ICI: 0.003 ICI: 0.001 ICI: 0.002 0.075 MCE: 0.003 0.15 MCE: 0.009 MCE: 0.012 MCE: 0.013 MCE: 0.010 0.075 MCE: 0.011 0.04 0.10 0.050 0.10 0.050 0.1 0.02 0.05 0.025 0.05 0.025 0.00 0.000 0.00 0.0 0.00 0.000 0.0 0.1 0.00 0.05 0.0 0.1 0.0 0.1 0.2 0.000 0.025 0.050 0.00 0.05 Venous Thromboembolism Retinal Detachment Chronic Otitis Media (n=92,077, I=2.0%, (n=91,313, I=0.1%, Multiple Sclerosis (n=90,662, I=1.0%, Migraine (n=90,133, Melanoma (n=88,489, 365d) 365d) (n=91,280, I=0.1%, 5y) 365d) I=2.3%, 365d) I=0.4%, 5y) 0.06 0.10 ICI: 0.002 ICI: 0.001 ICI: 0.001 ICI: 0.001 ICI: 0.001 ICI: 0.002 0.06 0.075 MCE: 0.003 MCE: 0.009 MCE: 0.003 MCE: 0.003 0.075 MCE: 0.014 0.10 MCE: 0.009 0.04 0.04 0.050 0.050 0.05 0.05 0.02 0.02 0.025 0.025 0.00 0.00 0.00 0.000 0.00 0.000 0.00 0.05 0.10 0.000 0.025 0.050 0.000 0.025 0.050 0.00 0.05 0.00 0.05 0.10 0.00 0.05 Predicted risk Predicted risk Predicted risk Predicted risk Predicted risk Predicted risk

Extended Data Figure 9: Calibration curves for new disease onset tasks. Predicted and observed risks are plotted in 20 bins, tasks are sorted by number of events (n) in the test set per task. To quantify the calibration performance, integrated calibration index (ICI) and mean calibration error (MCE) are given for every task; n: number of events; I: label imbalance; τ : task duration. 42

Observed risk

Meniere Disease Sarcoidosis (n=87,733, Hip Fracture (n=86,501, Peripheral Neuropathy Non-Hodgkin Lymphoma Hyperprolactinemia (n=88,001, I=0.1%, 3y) I=0.1%, 5y) I=0.7%, 365d) (n=85,988, I=1.3%, 3y) (n=85,695, I=0.3%, 5y) (n=83,097, I=0.2%, 3y) 0.075 ICI: 0.002 ICI: 0.001 0.001 ICI: 0.002 0.004 0.075 ICI: 0.001 0.10 ICI: 0.075 ICI: MCE: 0.006 MCE: 0.019 MCE: 0.011 MCE: 0.028 MCE: 0.013 0.10 MCE: 0.005 0.04 0.050 0.050 0.050 0.05 0.05 0.02 0.025 0.025 0.025 0.00 0.00 0.02 0.04

0.000 0.00

0.05

0.000 0.00

ICI: 0.003

0.075 MCE: 0.015 0.050 0.025 0.000 0.00 0.05

ICI: 0.001

0.075 MCE: 0.003 0.050 0.025 0.000 0.00 0.05

0.10

Cirrhosis (n=82,336, I=0.5%, 5y) ICI: 0.002 MCE: 0.013

0.05

0.10

Stroke (n=81,579, I=1.7%, 365d) 0.10 ICI: 0.002 MCE: 0.008

0.000 0.05 0.10 0.00 0.05 Primary Hyperparathyroidism Kidney Cancer (n=80,008, (n=80,549, I=0.2%, 5y) I=0.3%, 5y) ICI: 0.001 ICI: 0.002 0.06 MCE: 0.006 0.075 MCE: 0.005 0.04

0.05

0.00 0.00

0.050 0.025 0.000 0.00

0.02

0.00 0.00 0.00 0.00 0.05 0.10 0.00 0.05 0.10 0.000 0.025 0.050 0.05 All-Cause Mortality Interstitial Lung Glaucoma (n=79,639, Hidradenitis Suppurativa (n=78,240, I=7.7%, Disease (n=76,277, Type 2 Diabetes Liver Cancer (n=74,739, I=2.0%, 5y) (n=79,200, I=0.2%, 3y) 365d) I=0.7%, 5y) (n=76,085, I=3.4%, 3y) I=0.2%, 5y) 0.004 ICI: 0.002 ICI: 0.005 ICI: 0.004 ICI: 0.007 ICI: 0.002 0.075 0.4 0.15 ICI: 0.075 MCE: 0.008 0.3 MCE: 0.022 MCE: 0.022 MCE: 0.025 MCE: 0.029 MCE: 0.031 0.10 0.050 0.10 0.050 0.2 0.2 0.05 0.025 0.05 0.025 0.1

0.00 0.000 0.0 0.00 0.0 0.000 0.0 0.1 0.00 0.05 0.0 0.2 0.4 0.00 0.05 0.10 0.0 0.2 0.00 0.05 Acute Myocardial Infarction (n=74,297, Non-Melanoma Skin Cancer Multiple Myeloma Aortic Aneurysm Chronic Kidney Disease Pancreatic Cancer I=1.1%, 365d) (n=73,798, I=1.9%, 3y) (n=71,663, I=0.2%, 5y) (n=70,080, I=1.0%, 5y) (n=69,163, I=4.7%, 3y) (n=69,000, I=0.3%, 2y) ICI: 0.001 0.001 ICI: 0.002 ICI: 0.003 0.006 ICI: 0.002 0.4 ICI: 0.15 ICI: MCE: 0.006 MCE: 0.021 0.15 MCE: 0.032 0.06 MCE: 0.009 0.06 MCE: 0.015 0.075 MCE: 0.014 0.10 0.10 0.04 0.04 0.050 0.2 0.05 0.05 0.02 0.02 0.025 0.000 0.00 0.00 0.00 0.0 0.00 0.00 0.05 0.0 0.1 0.00 0.05 0.0 0.1 0.0 0.2 0.4 0.00 0.05 Chronic Obstructive Chronic Lymphocytic Pulmonary Disease Leukemia (n=68,605, Aortic Stenosis Lung Cancer (n=67,433, Osteoarthritis Dyslipidemia (n=66,743, (n=68,687, I=1.5%, 5y) I=0.1%, 5y) (n=67,486, I=1.1%, 5y) I=0.9%, 6y) (n=67,258, I=6.6%, 5y) I=13.7%, 3y) 0.2 ICI: 0.006 ICI: 0.005 ICI: 0.003 ICI: 0.008 ICI: 0.008 0.2 ICI: 0.006 0.4 0.075 MCE: 0.031 MCE: 0.041 MCE: 0.045 MCE: 0.024 MCE: 0.018 0.2 MCE: 0.016 0.4 0.050 0.1 0.1 0.2 0.1 0.2 0.025 0.0 0.000 0.0 0.0 0.0 0.0 0.0 0.1 0.2 0.00 0.05 0.0 0.1 0.2 0.0 0.1 0.2 0.0 0.2 0.4 0.0 0.2 0.4 Peripheral Artery Chronic Ischemic Heart Osteoporosis (n=66,705, Heart Failure (n=66,617, Disease (n=65,448, Disease (n=65,113, Atrial Fibrillation Bladder Cancer I=3.5%, 5y) I=3.5%, 3y) I=1.4%, 5y) I=4.9%, 5y) (n=64,564, I=5.0%, 5y) (n=64,368, I=0.4%, 5y) 0.4 ICI: 0.005 0.3 ICI: 0.009 ICI: 0.004 0.006 0.009 0.003 0.10 ICI: 0.2 ICI: 0.4 ICI: MCE: 0.038 MCE: 0.032 MCE: 0.017 MCE: 0.026 MCE: 0.021 0.3 MCE: 0.024 0.2 0.2 0.2 0.05 0.1 0.2 0.1 0.1 0.0 0.0 0.0 0.0 0.0 0.00 0.0 0.2 0.0 0.2 0.0 0.1 0.2 0.0 0.2 0.4 0.0 0.2 0.4 0.00 0.05 0.10 Parkinson's Disease Hypertension (n=61,086, Macular Degeneration Ovarian Cancer Breast Cancer (n=48,093, Alzheimer's Disease (n=61,132, I=0.5%, 5y) I=10.7%, 3y) (n=60,845, I=2.5%, 5y) (n=50,889, I=0.2%, 5y) I=2.2%, 5y) (n=44,219, I=0.8%, 5y) 0.15 ICI: 0.006 ICI: 0.015 0.005 0.002 ICI: 0.007 0.006 0.2 ICI: 0.075 ICI: 0.15 ICI: MCE: 0.059 MCE: 0.033 MCE: 0.013 MCE: 0.014 MCE: 0.027 0.2 MCE: 0.016 0.4 0.10 0.050 0.10 0.1 0.1 0.2 0.05 0.025 0.05 0.00 0.0 0.0 0.000 0.0 0.00 0.0 0.1 0.0 0.2 0.4 0.0 0.1 0.2 0.00 0.05 0.0 0.1 0.2 0.0 0.1 Polycystic Ovary Benign Prostatic Endometriosis (n=43,788, Endometrial Cancer Syndrome (n=28,485, Hyperplasia (n=25,344, Prostate Cancer I=0.8%, 3y) (n=41,462, I=0.4%, 5y) I=1.1%, 3y) I=6.7%, 3y) (n=24,948, I=3.1%, 5y) 0.3 ICI: 0.009 ICI: 0.004 ICI: 0.005 0.005 0.008 0.3 ICI: 0.15 ICI: 0.10 MCE: 0.041 MCE: 0.023 MCE: 0.028 MCE: 0.030 0.10 MCE: 0.028 0.2 0.2 0.10 0.05 0.05 0.1 0.1 0.05

Observed risk

Colorectal Cancer Gout (n=82,453, I=1.1%, (n=82,952, I=0.3%, 5y) 365d)

0.00 0.00

0.05

0.00 0.00

0.05 0.10 Predicted risk

0.00 0.00

0.05 0.10 Predicted risk

0.00 0.0

0.0 0.0

0.1 Predicted risk

0.2 Predicted risk

0.0 0.0

0.2 Predicted risk

Extended Data Figure 10: Calibration curves for new disease onset tasks. Continued. 43

0.2

ICI: 0.005 MCE: 0.024

Hypertension to Heart Failure (n=59,782, I=6.1%, 5y) 0.2

ICI: 0.004 MCE: 0.016

0.1

0.0 0.0 0.1 0.2 Diabetes Mellitus to Peripheral Artery Disease (n=22,170, I=3.3%, 5y)

0.0 0.0

0.1

ICI: 0.003 MCE: 0.010

0.2

0.1

0.2

GERD to Barrett's Esophagus (n=29,798, I=1.5%, 5y)

ICI: 0.005 MCE: 0.020

ICI: 0.004 MCE: 0.019

0.05

0.1 0.0 0.0

0.1

0.2

Obesity to Type 2 Diabetes (n=24,164, I=9.1%, 5y) 0.2

ICI: 0.010 MCE: 0.024

0.1

0.00 0.00

0.05

0.10

ICI: 0.005 MCE: 0.018

ICI: 0.003 MCE: 0.011

0.05

0.1

0.00 0.00 0.05 0.10

ICI: 0.004 MCE: 0.010

0.0 0.0

0.00 0.00

0.10

Asthma to Exacerbation Requiring Oral Corticosteroids (n=24,162, I=7.8%, 3y) 0.2 ICI: 0.005 MCE: 0.024

0.1

0.2

Diabetes Mellitus to Type 2 Diabetes to Type 2 Diabetes to Type 2 Diabetes to Diabetic Retinopathy Diabetic Foot Ulcer Diabetic Nephropathy Diabetic Neuropathy (n=21,954, I=4.4%, 5y) (n=21,465, I=0.9%, 5y) (n=21,252, I=1.6%, 5y) (n=20,438, I=5.0%, 5y) 0.05

0.0 0.0

Hypertension to Atrial Fibrillation (n=57,973, I=8.1%, 5y)

ICI: 0.005 MCE: 0.013

0.05

0.0 0.0

0.1

0.2

Type 2 Diabetes to Insulin Use (n=19,528, I=23.0%, 3y) 0.2

ICI: 0.010 MCE: 0.031

0.00 0.00 0.0 0.00 0.05 0.00 0.05 0.10 0.0 0.2 Chronic Ischemic Heart Chronic Ischemic Heart Episodic to Chronic Disease to Acute Disease to Heart Failure Atrial Fibrillation to Migraine (n=13,503, NAFLD to Liver Fibrosis NAFLD to NASH (n=12,761, Myocardial Infarction (n=12,096, I=12.6%, Ischemic Stroke I=0.8%, 365d) (n=12,813, I=0.4%, 5y) I=0.7%, 3y) (n=12,530, I=4.5%, 3y) 3y) (n=10,916, I=3.6%, 3y) ICI: 0.001 ICI: 0.003 ICI: 0.003 ICI: 0.007 ICI: 0.008 ICI: 0.007 0.10 MCE: 0.019 MCE: 0.021 MCE: 0.051 0.050 MCE: 0.011 0.050 MCE: 0.011 0.050 MCE: 0.009 0.1 0.2 0.05 0.025 0.025 0.025 0.05

0.000 0.000 0.000 0.000 0.025 0.050 0.000 0.025 0.050 0.000 0.025 0.050

0.00 0.0 0.0 0.00 0.05 0.10 0.0 0.2 0.0 0.1 Inflammatory Bowel Benign Prostatic COVID-19 Mortality Osteoporosis to Bone Disease to Colorectal Acute COVID-19 to Long Hyperplasia to Urinary (n=10,165, I=0.5%, Fracture (n=10,084, Glaucoma to Blindness Cancer (n=8,794, I=0.5%, COVID (n=8,165, I=0.7%, Retention (n=7,929, 60d) I=1.2%, 3y) (n=9,408, I=1.1%, 5y) 5y) 365d) I=4.9%, 3y) 0.10 ICI: 0.001 ICI: 0.003 ICI: 0.008 ICI: 0.004 ICI: 0.006 ICI: 0.008 MCE: 0.006 MCE: 0.018 MCE: 0.070 MCE: 0.040 MCE: 0.039 MCE: 0.052 0.10 0.10 0.04 0.05 0.05 0.05 0.05 0.05 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.04 0.00 0.05 0.00 0.05 0.10 0.00 0.05 0.00 0.05 0.10 0.00 0.05 0.10 Prediabetes to Type 2 Breast Cancer Palliative Heart Failure to Acute Diabetes (n=7,776, Heart Failure Mortality Breast Cancer Mortality Care (n=6,733, I=2.9%, Breast Cancer Metastasis Kidney Failure (n=6,354, I=6.2%, 3y) (n=6,806, I=29.8%, 3y) (n=6,736, I=12.7%, 3y) 5y) (n=6,514, I=4.4%, 5y) I=9.6%, 365d) ICI: 0.011 ICI: 0.019 ICI: 0.009 ICI: 0.013 ICI: 0.013 ICI: 0.008 0.4 MCE: 0.025 MCE: 0.022 0.2 MCE: 0.067 0.2 MCE: 0.124 0.2 MCE: 0.067 0.50 MCE: 0.063 0.1 0.2 0.1 0.1 0.1 0.25 0.0 0.0

0.00 0.0 0.0 0.0 0.0 0.1 0.2 0.00 0.25 0.50 0.0 0.2 0.4 0.0 0.1 0.2 0.0 0.1 0.2 0.0 0.1 COPD to Chronic COPD to Pulmonary Chronic Hepatitis or Respiratory Failure Hypertension (n=5,788, Fatty Liver to Cirrhosis CKD Stage 3 to Stage 4+ COPD to Exacerbation Stroke Mortality (n=5,858, I=2.8%, 3y) I=3.0%, 3y) (n=5,786, I=3.4%, 5y) (n=5,442, I=4.6%, 3y) (n=5,425, I=5.5%, 3y) (n=5,239, I=18.8%, 3y) 0.2 ICI: 0.008 ICI: 0.008 ICI: 0.015 ICI: 0.010 ICI: 0.009 ICI: 0.011 0.2 MCE: 0.052 MCE: 0.039 MCE: 0.045 MCE: 0.123 MCE: 0.052 0.50 MCE: 0.044 0.10 0.1 0.2 0.1 0.1 0.25 0.05 0.0 0.00 0.0 0.1 0.00 0.05 0.10 Barrett's Esophagus to Prostate Cancer Esophageal Cancer Mortality (n=4,533, (n=5,018, I=1.0%, 5y) I=16.2%, 5y) ICI: 0.004 MCE: 0.016

0.05

0.2

0.50

ICI: 0.022 MCE: 0.063

0.0 0.0

0.2 Prostate Cancer Metastasis (n=4,426, I=3.5%, 5y)

0.2

ICI: 0.021 MCE: 0.155

0.25

0.00 0.00

Observed risk

Hypertension to Chronic Kidney Disease Stage 3+ (n=60,041, I=6.9%, 5y)

0.0 0.0 0.00 0.0 0.1 0.2 0.0 0.1 0.2 0.00 0.25 0.50 Paroxysmal to Persistent Thyroid Cancer Psoriasis to Psoriatic Afib (n=4,176, I=2.8%, Metastasis (n=3,701, Arthritis (n=3,521, 3y) I=2.7%, 5y) I=4.5%, 5y) ICI: 0.015 ICI: 0.010 ICI: 0.021 0.2 0.2 MCE: 0.102 MCE: 0.075 MCE: 0.108 0.1

0.1

0.00 0.0 0.0 0.0 0.0 0.00 0.25 0.50 0.0 0.2 0.0 0.1 0.0 0.1 0.2 0.0 0.1 0.2 Non-Proliferative to Rheumatoid Arthritis to Bladder Cancer Proliferative Diabetic Interstitial Lung Rheumatoid Arthritis to Metastasis (n=3,491, Retinopathy (n=2,921, Lung Cancer Mortality Disease (n=2,686, Second-Line Therapy Lung Cancer Metastasis I=3.9%, 3y) I=3.7%, 3y) (n=2,701, I=39.2%, 2y) I=2.3%, 5y) (n=2,681, I=6.5%, 3y) (n=2,534, I=9.0%, 2y) 0.2 ICI: 0.014 ICI: 0.015 ICI: 0.054 ICI: 0.013 ICI: 0.011 ICI: 0.023 0.05

MCE: 0.094

0.1 0.0 0.0

MCE: 0.050

0.1 0.1 0.2 Predicted risk

0.0 0.0

0.1 Predicted risk

0.5

MCE: 0.142

0.0 0.0

0.5 Predicted risk

0.1

MCE: 0.068

0.0 0.0

MCE: 0.043

0.2

0.1 0.1 Predicted risk

0.0 0.0

MCE: 0.068

0.1 0.2 Predicted risk

0.0 0.0

0.2 Predicted risk

Extended Data Figure 11: Calibration curves for disease progression tasks. Predicted and observed risks are plotted in 20 bins, tasks are sorted by number of events (n) in the test set per task. To quantify the calibration performance, integrated calibration index (ICI) and mean calibration error (MCE) are given for every task; n: number of events; I: label imbalance; τ : task duration. 44

Observed risk

Aortic Stenosis Mortality (n=2,449, I=23.4%, 3y)

Ovarian Cancer Colorectal Cancer Melanoma Mortality Metastasis (n=1,967, Metastasis (n=1,913, (n=1,974, I=13.4%, 3y) I=4.9%, 3y) I=6.5%, 3y) 0.040 0.033 ICI: 0.041 ICI: 0.022 ICI: 0.026 ICI: 0.039 0.2 ICI: 0.6 ICI: 0.2 MCE: 0.113 MCE: 0.110 MCE: 0.145 MCE: 0.069 MCE: 0.128 MCE: 0.186 0.3 0.6 0.4 0.4 0.2 0.4 0.1 0.1 0.2 0.2 0.1 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.00 0.25 0.50 0.0 0.1 0.2 0.0 0.5 0.0 0.2 0.4 0.0 0.1 0.2 0.0 0.2 Pulmonary Fibrosis to Cirrhosis to Liver Melanoma Metastasis Hepatitis C to Cirrhosis All Leukemia Mortality Liver Cancer Metastasis Pulmonary Hypertension Cancer (n=1,293, I=3.7%, (n=1,905, I=5.2%, 2y) (n=1,601, I=6.6%, 5y) (n=1,451, I=25.2%, 3y) (n=1,404, I=3.6%, 2y) (n=1,369, I=6.2%, 3y) 3y) 0.3 ICI: 0.020 0.3 ICI: 0.021 ICI: 0.035 ICI: 0.022 ICI: 0.032 0.3 ICI: 0.040 MCE: 0.117 MCE: 0.142 MCE: 0.074 MCE: 0.078 MCE: 0.077 MCE: 0.209 0.2 0.6 0.2 0.2 0.2 0.2 0.4 0.1 0.1 0.1 0.1 0.1 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.2 0.0 0.2 0.0 0.5 0.0 0.1 0.2 0.0 0.1 0.2 0.0 0.2 Cirrhosis to Esophageal Myelodysplastic Syndrome Gestational Diabetes to Varices (n=1,252, to Leukemia (n=1,248, Thyroid Cancer Mortality Bladder Cancer Mortality Type 2 Diabetes Kidney Cancer Mortality I=6.9%, 3y) I=8.5%, 2y) (n=1,246, I=7.9%, 5y) (n=1,187, I=25.2%, 3y) (n=1,150, I=7.2%, 5y) (n=869, I=20.8%, 3y) 0.6 ICI: 0.025 ICI: 0.029 ICI: 0.050 0.042 0.75 ICI: 0.052 0.3 ICI: 0.032 0.3 ICI: MCE: 0.074 MCE: 0.256 MCE: 0.183 MCE: 0.213 0.6 MCE: 0.149 0.3 MCE: 0.079 0.4 0.50 0.2 0.2 0.4 0.2 0.2 0.25 0.1 0.1 0.2 0.1 0.0 0.0

0.2

Observed risk

0.0 0.0

0.2

Colorectal Cancer Mortality (n=2,053, I=23.9%, 3y)

0.0 0.00

0.25

0.0 0.00 0.25 0.50

0.50

0.0 0.0

0.2 Mild Cognitive Impairment to Alzheimer's Disease (n=698, I=6.9%, 3y)

0.00 0.0

0.5

Pancreatic Cancer Mild Cognitive CKD Stage 2 to Stage 3 Mortality (n=766, Impairment to Dementia (n=794, I=17.1%, 5y) I=50.8%, 2y) (n=674, I=12.0%, 3y) 0.6 ICI: 0.084 ICI: 0.022 ICI: 0.039 0.082 ICI: 0.107 0.097 0.6 ICI: 0.6 ICI: 0.4 MCE: 0.117 MCE: 0.343 MCE: 0.628 MCE: 0.474 0.75 MCE: 0.255 0.2 MCE: 0.134 0.4 0.4 0.4 0.50 0.2 0.1 0.2 0.2 0.2 0.25 0.0 0.0 0.0 0.00 0.0 0.0 0.0 0.1 0.2 0.0 0.2 0.4 0.00 0.25 0.50 0.0 0.5 0.00 0.25 0.50 0.00 0.25 0.50 Chronic Lymphocytic Chronic Myeloid Leukemia Acute Lymphocytic Acute Myeloid Leukemia Ovarian Cancer Mortality Leukemia Mortality Mortality (n=492, Liver Cancer Mortality Leukemia Mortality Mortality (n=343, (n=670, I=31.9%, 3y) (n=618, I=18.0%, 5y) I=16.7%, 5y) (n=472, I=40.3%, 365d) (n=412, I=21.8%, 2y) I=42.9%, 2y) 0.6 ICI: 0.067 0.75 ICI: 0.069 ICI: 0.099 ICI: 0.075 ICI: 0.142 ICI: 0.130 0.75 0.75 MCE: 0.252 MCE: 0.271 MCE: 0.450 MCE: 0.248 0.75 MCE: 0.403 0.6 MCE: 0.204 0.4 0.50 0.50 0.50 0.50 0.4 0.2 0.25 0.25 0.25 0.25 0.2 0.0 0.00 0.00 0.00 0.0 0.00 0.0 0.5 0.0 0.5 0.0 0.5 0.0 0.5 0.00 0.25 0.50 0.0 0.5 Predicted risk Predicted risk Predicted risk Predicted risk Predicted risk Predicted risk NASH to Cirrhosis (n=852, I=6.5%, 3y)

Observed risk

Pancreatic Cancer Metastasis (n=2,169, I=4.8%, 2y)

Kidney Cancer Metastasis (n=826, I=8.7%, 2y)

Extended Data Figure 12: Calibration curves for disease progression tasks. Continued.

ICI: 0.029 MCE: 0.088

MACE-Free Survival on MACE-Free Survival on Lisinopril for Aspirin for Chronic Survival on Furosemide Hypertension (n=4,986, Ischemic Heart Disease for Heart Failure I=20.1%, 3y) (n=3,491, I=26.8%, 3y) (n=2,779, I=34.7%, 2y) ICI: 0.027 MCE: 0.092

ICI: 0.034 MCE: 0.094

ICI: 0.033 MCE: 0.074

MACE-Free Survival on MACE-Free Survival on Losartan for Atorvastatin for Chronic Hypertension (n=2,738, Ischemic Heart Disease I=18.3%, 3y) (n=2,558, I=28.1%, 3y) ICI: 0.028 MCE: 0.111

ICI: 0.043 MCE: 0.112

0.50 0.50 0.5 0.5 0.5 0.25 0.25 0.25 0.00 0.00 0.0 0.00 0.0 0.0 0.00 0.25 0.50 0.00 0.25 0.50 0.0 0.5 0.00 0.25 0.50 0.0 0.5 0.0 0.5 Survival on Metoprolol Stroke/Embolism-Free ESRD-Free Survival on MACE-Free Survival on HF-Free Survival on Succinate for Heart Survival on Apixaban for Lisinopril for Chronic Clopidogrel for Chronic Survival on Lisinopril Empagliflozin for Type 2 Failure (n=2,427, Atrial Fibrillation Kidney Disease (n=1,028, Ischemic Heart Disease for Heart Failure Diabetes (n=972, I=34.4%, 2y) (n=2,007, I=23.4%, 3y) I=27.2%, 3y) (n=1,006, I=27.0%, 3y) (n=981, I=31.8%, 2y) I=14.2%, 365d) ICI: 0.032 ICI: 0.056 ICI: 0.062 ICI: 0.084 ICI: 0.054 0.036 0.50 ICI: MCE: 0.089 MCE: 0.159 MCE: 0.150 MCE: 0.092 0.50 MCE: 0.205 0.50 MCE: 0.212 0.50 0.5 0.5 0.25 0.25 0.25 0.25 0.50

0.00 0.00 0.25 0.50

0.0 0.0 0.00 0.00 0.00 0.0 0.5 0.0 0.5 0.00 0.25 0.50 0.00 0.25 0.50 0.00 0.25 0.50 Stroke/Embolism-Free Survival on Stroke/Embolism-Free MACE-Free Survival on Survival on Trastuzumab Survival on Warfarin for Spironolactone for Heart Survival on Leuprolide Survival on Rivaroxaban Ticagrelor for Chronic for Breast Cancer Atrial Fibrillation Failure (n=893, I=30.1%, for Prostate Cancer for Atrial Fibrillation Ischemic Heart Disease (n=923, I=13.2%, 3y) (n=895, I=38.4%, 3y) 2y) (n=768, I=20.7%, 2y) (n=568, I=25.0%, 3y) (n=544, I=18.6%, 3y) ICI: 0.052 ICI: 0.060 ICI: 0.073 0.033 ICI: 0.087 ICI: 0.070 0.50 ICI: MCE: 0.201 MCE: 0.171 MCE: 0.206 MCE: 0.097 MCE: 0.280 MCE: 0.198 0.5 0.5 0.5 0.5 0.5 0.25 0.0 0.0 0.0 0.00 0.0 0.0 0.0 0.5 0.0 0.5 0.0 0.5 0.00 0.25 0.50 0.0 0.5 0.0 0.5 Exacerbation-Free Survival on Tiotropium Survival on Cisplatin Survival on Paclitaxel Survival on Carboplatin Survival on Pertuzumab Survival on Rituximab for COPD (n=542, for Head & Neck Cancer for Breast Cancer for Lung Cancer (n=446, for Breast Cancer for Non-Hodgkin Lymphoma I=40.4%, 3y) (n=499, I=17.4%, 2y) (n=490, I=17.3%, 3y) I=56.3%, 365d) (n=432, I=8.3%, 3y) (n=408, I=23.0%, 3y) 1.0 ICI: 0.070 ICI: 0.091 0.050 ICI: 0.117 ICI: 0.041 ICI: 0.106 0.50 ICI: MCE: 0.188 MCE: 0.249 MCE: 0.324 MCE: 0.280 MCE: 0.276 0.50 MCE: 0.318 0.5 0.5 0.5 0.5 0.25 0.25 0.0 0.0

0.00 0.0 0.0 0.00 0.0 0.00 0.25 0.50 0.0 0.5 1.0 0.0 0.5 0.00 0.25 0.50 0.0 0.5 MACE-Free Survival on Survival on Sacubitril / Survival on Survival on Fluorouracil Chlorthalidone for Survival on Carboplatin Valsartan for Heart Survival on Nivolumab Pembrolizumab for Lung for Colorectal Cancer Hypertension (n=344, for Endometrial Cancer Failure (n=324, I=17.3%, for Melanoma (n=304, Cancer (n=252, I=57.1%, (n=360, I=41.4%, 2y) I=17.7%, 3y) (n=338, I=30.2%, 2y) 2y) I=29.3%, 2y) 365d)

0.5

ICI: 0.140 MCE: 0.294

0.5

ICI: 0.107 MCE: 0.403

0.5

ICI: 0.145 MCE: 0.412

Observed risk

ICI: 0.122 MCE: 0.385

0.5

ICI: 0.110 MCE: 0.255

0.5

0.50

ICI: 0.132 MCE: 0.304

ICI: 0.141 MCE: 0.399

0.5

0.25

0.0 0.0 0.5 0.0 0.5 0.0 0.5 Survival on Survival on Lactulose Survival on Pembrolizumab for Breast for Hepatic Pembrolizumab for Cancer (n=247, I=21.1%, Encephalopathy (n=243, Melanoma (n=235, 3y) I=30.9%, 2y) I=28.9%, 2y)

0.5

Observed risk

0.5

0.0 0.0

Observed risk

MACE-Free Survival on Amlodipine for Hypertension (n=5,059, I=21.5%, 3y)

ICI: 0.164 MCE: 0.538

0.00 0.00

ICI: 0.109 MCE: 0.430

0.5

0.0 0.0 0.0 0.5 0.0 0.5 Exacerbation-Free Survival on Rituximab Survival on Carboplatin Survival on Umeclidinium for Diffuse Large B-Cell for Ovarian Cancer / Vilanterol for COPD Lymphoma (n=214, (n=225, I=42.2%, 2y) (n=222, I=31.1%, 3y) I=25.2%, 2y) 1.0 ICI: 0.148 ICI: 0.153 ICI: 0.141

0.5

0.25

0.50

MCE: 0.417

MCE: 0.408

MCE: 0.524

0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.5 0.0 0.5 0.0 0.5 0.0 0.5 1.0 0.0 0.5 Survival on Cytarabine for Acute Myeloid Survival on Paclitaxel Survival on Oxaliplatin Survival on Gemcitabine Survival on Cisplatin Survival on Carboplatin Leukemia (n=210, for Ovarian Cancer for Stomach Cancer for Pancreatic Cancer for Bladder Cancer for Esophageal Cancer I=41.4%, 2y) (n=207, I=38.2%, 2y) (n=196, I=52.0%, 2y) (n=191, I=68.1%, 365d) (n=188, I=26.6%, 2y) (n=187, I=41.2%, 2y) 1.0 1.0 ICI: 0.208 1.0 ICI: 0.122 1.0 ICI: 0.218 ICI: 0.152 ICI: 0.215 ICI: 0.140 MCE: 0.645

MCE: 0.218

0.5

MCE: 0.740

MCE: 0.416

0.5

MCE: 0.391

MCE: 0.637

0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 0.0 0.5 1.0 Survival on Sildenafil Survival on Octreotide for Pulmonary Arterial Survival on Doxorubicin Survival on Bortezomib Survival on Paclitaxel Survival on Atezolizumab for Neuroendocrine Hypertension (n=187, for Soft Tissue Sarcoma for Multiple Myeloma for Pancreatic Cancer for Lung Cancer (n=159, Tumors (n=148, I=31.1%, I=37.4%, 3y) (n=181, I=55.2%, 2y) (n=171, I=31.6%, 2y) (n=161, I=70.8%, 365d) I=62.3%, 365d) 3y) 1.0 ICI: 0.130 1.0 ICI: 0.202 1.0 ICI: 0.229 ICI: 0.203 ICI: 0.163 ICI: 0.207 MCE: 0.345

0.0 0.0

MCE: 0.453

0.5

0.5 0.5 1.0 Predicted risk

0.0 0.0

0.5 Predicted risk

0.5

MCE: 0.446

0.0 0.0

MCE: 0.565

0.5 0.5 Predicted risk

0.0 0.0

MCE: 0.631

0.5 0.5 1.0 Predicted risk

0.0 0.0

0.5 1.0 Predicted risk

0.5

MCE: 0.616

0.0 0.0

0.5 Predicted risk

Extended Data Figure 13: Calibration curves for treatment response tasks. Predicted and observed risks are plotted in 20 bins, tasks are sorted by number of events (n) in the test set per task. To quantify the calibration performance, integrated calibration index (ICI) and mean calibration error (MCE) are given for every task; n: number of events; I: label imbalance; τ : task duration. 46

Observed risk Observed risk

Survival on Azacitidine Survival on Azacitidine Survival on Riluzole for Survival on Nivolumab for Acute Myeloid Survival on Abiraterone Survival on Temozolomide for Myelodysplastic Amyotrophic Lateral for Kidney Cancer Leukemia (n=111, for Prostate Cancer for Glioblastoma (n=100, Syndrome (n=99, I=56.6%, Sclerosis (n=95, (n=140, I=45.0%, 2y) I=57.7%, 2y) (n=100, I=53.0%, 2y) I=71.0%, 365d) 2y) I=45.3%, 2y) 1.0 ICI: 0.260 1.0 ICI: 0.237 1.0 ICI: 0.229 1.0 ICI: 0.323 1.0 ICI: 0.308 1.0 ICI: 0.291 MCE: 0.727

MCE: 0.846

0.5

MCE: 0.736

0.5

0.0 0.0

MCE: 0.655

0.5

MCE: 0.561

MCE: 0.904

0.5

0.0 0.0 0.0 0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0 0.0 Survival on Survival on Enfortumab Survival on Ipilimumab Pembrolizumab for Survival on Osimertinib Survival on Pemetrexed vedotin for Bladder for Kidney Cancer (n=84, Colorectal Cancer (n=81, for Lung Cancer (n=79, for Mesothelioma (n=75, Cancer (n=72, I=36.1%, I=41.7%, 2y) I=43.2%, 2y) I=55.7%, 365d) I=60.0%, 2y) 2y) 1.0 1.0 ICI: 0.239 1.0 ICI: 0.250 1.0 ICI: 0.238 1.0 ICI: 0.211 0.5

1.0

0.0 0.0

MCE: 0.622

MCE: 0.731

0.5

MCE: 0.588

0.5

0.0 0.0

0.5 1.0 Predicted risk

0.5

0.0 0.0

1.0

MCE: 0.614

0.5

0.0 0.0

0.5

0.5 1.0 Predicted risk

0.0 0.0

0.5 1.0 Predicted risk

0.0

0.5 1.0 Predicted risk

Extended Data Figure 14: Calibration curves for treatment response tasks. Continued.

Observed risk

Heparin-Induced Thrombocytopenia (n=112,549, I=0.1%, 14d) ICI: 0.000

ICI: 0.000

0.04 MCE: 0.001

0.04 MCE: 0.002

0.02

0.00 0.00

0.02

0.04

0.00 0.00

0.02

0.04

Acute Kidney Injury After NSAIDs (n=65,679, I=1.8%, 30d) 0.05

ICI: 0.001 MCE: 0.008

Tardive Dyskinesia on Type 2 Diabetes After Antipsychotics Systemic Corticosteroids (n=64,815, I=0.1%, 2y) (n=64,078, I=2.2%, 2y) ICI: 0.000

0.04 MCE: 0.001 0.02

0.1

ICI: 0.002 MCE: 0.014

Rhabdomyolysis on Statins (n=58,987, I=0.2%, 365d) ICI: 0.001

0.04 MCE: 0.003 0.02

0.00 0.00 0.00 0.0 0.00 0.00 0.02 0.04 0.00 0.05 0.00 0.02 0.04 0.0 0.1 0.00 0.02 0.04 GI Bleeding on Clostridioides difficile Angioedema on ACE Antiplatelet Agents Infection on PPIs Hyperkalemia on ACE Major Bleeding on Oral Type 2 Diabetes After Inhibitors (n=29,245, (n=26,571, I=1.7%, (n=22,134, I=0.8%, Inhibitors (n=9,647, Anticoagulants (n=8,387, Atypical Antipsychotics I=0.3%, 90d) 365d) 365d) I=2.4%, 30d) I=4.4%, 365d) (n=7,738, I=2.4%, 2y) 0.10 ICI: 0.003 ICI: 0.000 ICI: 0.003 ICI: 0.002 ICI: 0.008 ICI: 0.007 0.2 0.04 MCE: 0.000 MCE: 0.015 MCE: 0.017 MCE: 0.007 MCE: 0.041 MCE: 0.034 0.050 0.10 0.05 0.05 0.1 0.02 0.025 0.05

Hyperthyroidism on Amiodarone (n=7,323, I=0.7%, 2y) Observed risk

GI Bleeding on NSAIDs (n=66,042, I=0.5%, 90d)

ICI: 0.006

0.10 MCE: 0.063 0.05 0.00 0.00

0.05 0.10 Predicted risk

0.00 0.00

0.05

0.10

Cardiomyopathy After Anthracyclines (n=3,190, I=3.4%, 5y) 0.2

ICI: 0.019 MCE: 0.061

0.0 0.0

0.2 Predicted risk

0.00 0.00

0.05

0.000 0.000 0.025 0.050

0.00 0.0 0.00 0.05 0.10 0.0 MACE on Levothyroxine Hypothyroidism on with History of Heart Amiodarone (n=2,193, Failure (n=602, I=38.5%, I=3.9%, 2y) 2y)

Hypoglycemia on Sulfonylureas (n=2,918, I=2.4%, 365d) 0.05

ICI: 0.006 MCE: 0.020

0.00 0.00

0.2

ICI: 0.032 MCE: 0.155

0.0 0.0

0.05 Predicted risk

0.2 Predicted risk

0.5

0.1

0.2

ICI: 0.066 MCE: 0.125

0.0 0.0

0.5 Predicted risk

Extended Data Figure 15: Calibration curves for adverse events tasks. Predicted and observed risks are plotted in 20 bins, tasks are sorted by number of events (n) in the test set per task. To quantify the calibration performance, integrated calibration index (ICI) and mean calibration error (MCE) are given for every task; n: number of events; I: label imbalance; τ : task duration.

Observed risk

Acute Respiratory Distress Syndrome (n=219,603, I=0.0%, 6d) 0.04

ICI: 0.000 MCE: 0.000

0.02

Acute Clostridioides Acute Venous Acute Cardiac Arrest difficile Infection Acute Liver Failure Thromboembolism Acute Sepsis (n=59,932, (n=70,065, I=0.2%, 6d) (n=67,118, I=0.4%, 6d) (n=64,401, I=0.1%, 6d) (n=61,349, I=2.4%, 6d) I=1.4%, 6d) 0.04

ICI: 0.000 MCE: 0.001

0.02

0.04

ICI: 0.000 MCE: 0.002

0.04

ICI: 0.000 MCE: 0.001

ICI: 0.001 MCE: 0.002

0.025

0.02

0.050

0.04

ICI: 0.000 MCE: 0.003

0.02

0.00 0.00

0.00 0.00 0.00 0.000 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.02 0.04 0.000 0.025 0.050 0.00 0.02 0.04 Acute Dialysis Inpatient Length of Stay Acute CatheterInpatient Readmission Dependence (n=53,315, Acute Pressure Ulcer Acute Kidney Failure (n=45,508, I=100.0%, Associated UTI (All Cause) (n=13,718, I=0.4%, 6d) (n=51,579, I=0.7%, 6d) (n=51,353, I=4.4%, 6d) 6d) (n=44,623, I=0.1%, 6d) I=100.0%, 30d) 1.0 ICI: 0.018 ICI: 0.000 ICI: 0.000 ICI: 0.001 ICI: 0.000 ICI: 0.016 0.50 MCE: 0.008 MCE: 0.082 MCE: 0.030 0.04 MCE: 0.002 0.04 MCE: 0.000 0.04 MCE: 0.002 0.05 0.5 0.25 0.02 0.02 0.02 0.00 0.00 0.02 0.04 Predicted risk

0.00 0.00 0.02 0.04 Predicted risk

0.00 0.00

0.05 Predicted risk

0.0 0.0

0.5 1.0 Predicted risk

0.00 0.00

0.02 0.04 Predicted risk

0.00 0.00

0.25 0.50 Predicted risk

Extended Data Figure 16: Calibration curves for clinical operation management tasks. Predicted and observed risks are plotted in 20 bins, tasks are sorted by number of events (n) in the test set per task. To quantify the calibration performance, integrated calibration index (ICI) and mean calibration error (MCE) are given for every task; n: number of events; I: label imbalance; τ : task duration.

Extended Data Figure 17: Ablation study on architectural design choices of A POLLO, evaluated on disease progression of neoplastic diseases. Top-10 tasks are shown in Figure 3i, additional tasks shown here. Tasks are sorted by A POLLO ’s performance. n, number of patients (number of uncensored patients in parentheses); τ , time-to-event.

Extended Data Figure 18: Quantitative and qualitative retrieval performance. (a) 61 patient retrieval tasks curated from combinations of ICD10 diagnosis codes and medications, assessed with accuracy among the five closest (Acc@5) embedded patients compared to retrieval using the last progress note embedding. (b)-(c) Top 4 retrieved patients using external TCGA slides for invasive colon adenocarcinoma (b) and lung adenocarcinoma (c). To preserve patient privacy, ages shown in the figure reported as ranges instead of exact numbers.

Extended Data Figure 19: Additional local and global interpretability examples. (a) Local analysis. Three-year mortality risk trajectory for a patient with metastatic ROS1/ALK-altered lung adenocarcinoma receiving targeted therapy. The largest contributors in each interval correspond to staging and restaging reports, thoracic imaging, treatment-response assessments, new thromboembolic events (pulmonary embolism and deep-vein thrombosis), and functional descriptors such as pain score, weight and fatigue. (b–e) Global analysis. Population-level feature importance for four additional downstream tasks—ovarian cancer metastasis, cardiomyopathy after heart failure treated with anthracyclines, survival after aortic stenosis, and stroke/embolism-free survival with apixaban—computed via Integrated Gradients (IG) as in Fig. 5. For ovarian cancer metastasis (b), postprocedural states are consistent with the role of surgical and staging pathways in advanced ovarian cancer [65]. Breast-related terms such as benign mammary dysplasia plausibly act as a proxy for hereditary breast–ovarian cancer susceptibility and the downstream surveillance/diagnostic cascade, and BRCA1/2-associated HBOC confers markedly elevated ovarian and breast cancer risk [66]. Nicotine dependence is additionally linked to increased risk of mucinous ovarian cancer and has been associated with worse survival after ovarian cancer diagnosis [67]. For cardiomyopathy after heart failure treated with anthracyclines (c), the emergence of lymphoma and leukemia terms is clinically coherent because these diagnoses track anthracycline exposure and treatment intensity in real-world care. Anthracycline-related LV dysfunction is strongly influenced by cumulative dose and is amplified by patient-level risk factors such as older age, pre-existing cardiovascular disease, and cardiometabolic comorbidities [68]. The specific appearance of myeloid leukemia aligns with recent cardio-oncology literature emphasizing that older AML populations treated with anthracycline-containing regimens have elevated cardiotoxicity risk [68]. For survival after aortic stenosis (d), top attributors concentrate on multimorbidities including cognitive impairment and dementia, sensory loss, urinary incontinence, heart failure, venous thromboembolism, CKD, and intercurrent infection [69]. CKD is also a well-established modifier of prognosis in severe aortic stenosis and after TAVR, with higher mortality and complications as kidney disease advances. For apixaban-treated stroke/embolism-free survival (e), several high-importance features including blood-pressure strata, obesity/weight patterns, dementia reflect baseline thromboembolic risk and frailty. Benign prostatic hyperplasia (BPH) also appears as the top risk factor, which is found to be associated with higher rates of hematuria-related complications in patients exposed to antithrombotics [70]. Clinically, hematuria complications can drive reduced adherence or discontinuation of anticoagulation, where both bleeding-related discontinuation and non-persistence have been linked to higher risks of stroke/systemic embolism and death in AF populations [70]. To preserve patient privacy, ages shown in the figure are randomly shifted by -2 to 2 years from their actual age, and lab test and vitals values are presented as ranges instead of exact numbers.

Data Distributions Extended Data Table 1: Cohort demographics. Demographic

Category

Count

Sex

Male Female Unknown

3,915,625 3,236,747 2,672

Age (years)

0–17 18–34 35–49 50–64 65–79 80–89 90–99 ≥100

704,175 1,328,286 1,356,313 1,401,588 1,395,482 529,131 285,796 154,273

Total

7,155,044

Extended Data Table 2: Modality distribution of MGB-7M. Type

Modality

Detailed breakdown

Structured

Flowsheet Lab test Vital signs Medications Diagnosis

— — — — —

Clinical notes

Unstructured

Multimodal diagnostic reports

Imaging

Count 22,936,351,260 1,442,631,333 505,193,723 112,189,178 39,144,980

Progress notes

158,683,290

Problem comment and assessment and plan

19,779,901

Procedure note

1,286,329

History overview

1,226,140

Discharge plans

575,063

History and presentation note

517,542

Nursing note

228,038

Total number of clinical notes

182,296,303

Radiology Imaging

46,643,405

Pathology

12,935,526

Cardiology Diagnostics/Monitoring

8,424,343

Vascular Studies

1,546,342

Nuclear Medicine

1,142,903

Ophthalmology – Diagnostics

1,131,192

Endocrinology (Bone Density (DXA/DEXA))

960,634

Labs

579,034

Molecular / Genetics

276,423

Neurology & Sleep Diagnostics

214,112

Endoscopy / Procedural Diagnostics

130,922

Pulmonary Function Tests

129,706

Others

3,864,339

Total number of diagnostic reports

77,978,881

Hematology blood smears

522,190

Histopathology whole slide images

311,769

Gross images

165,837

Electron microscopy (EM) images

158,439

Total number of images Total

1,158,235 25,296,943,893

Extended Data Table 3: Data distribution of medications in MGB-7M. Count of medication records per ATC category in the MGB-7M, sorted by number of records per category. ATC Category

Count

Systemic hormonal preparations, excluding sex hormones and insulins 565,857 Antiparasitic products, insecticides and repellents 328,169 Genito urinary system and sex hormones 242,019 Alimentary tract and metabolism 235,168 Nervous system 181,774 Cardiovascular system 167,356 Dermatologicals 57,285 Various 30,181 Respiratory system 17,107 Antineoplastic and immunomodulating agents 9,770 Blood and blood forming organs 6,762 Antiinfective for systemic use 5,561 Sensory organs 5,285 Musculo-skeletal system 64 Other 110,336,820

Extended Data Table 4: Data distribution of diagnoses in MGB-7M. Count of records per ICD10 chapter. #

Chapter Name

1 2 3 4 5 6 7 8 9 10 11 12 13

Certain infectious and parasitic diseases 698,584 Neoplasms 1,489,284 Diseases of the blood and blood-forming organs 668,249 Endocrine, nutritional and metabolic diseases 3,513,738 Mental and behavioural disorders 2,372,097 Diseases of the nervous system 1,595,885 Diseases of the eye and adnexa 1,482,031 Diseases of the ear and mastoid process 572,722 Diseases of the circulatory system 2,943,805 Diseases of the respiratory system 1,821,748 Diseases of the digestive system 2,081,121 Diseases of the skin and subcutaneous tissue 1,441,923 Diseases of the musculoskeletal system and 3,671,049 connective tissue Diseases of the genitourinary system 1,892,284 Pregnancy, childbirth and the puerperium 439,951 Certain conditions originating in the perinatal 246,620 period Congenital malformations, deformations and 243,414 chromosomal abnormalities Symptoms, signs and abnormal clinical and lab- 5,375,320 oratory findings Injury, poisoning and certain other conse883,940 quences of external causes External causes of morbidity and mortality 127,703 Factors influencing health status and contact 5,451,213 with health services Codes for special purposes 132,299

14 15 16 17 18 19 20 21 22

Count

Training hyperparameters Hyperparameter

Value

Layers Heads Head activation Embedding dimension

12 12 GELU 768

Mask ratio

0.3

AdamW β Effective Batch size Total training iterations Warmup iterations Learning rate schedule Learning rate (start) Base encoder & embedding layers learning rate (post warmup) Projection layers & prediction heads learning rate (post warmup) Learning rate (final) Weight decay Automatic mixed precision

(0.9, 0.999) 1024 30,000 3,000 Cosine 0 2e-4 6e-4 0 1e-5 fp16

Extended Data Table 5: Hyperparameters used in pretraining A POLLO. 8 × 80GB NVIDIA A100 GPUs were used for training.

Downstream evaluation Extended Data Table 6: Definition of new disease onset prediction tasks. We evaluated TTE performance on 95 disease onset tasks. τ is the threshold duration. The snapshot is taken at time of random hospital discharge. For certain sex-specific diseases, we filter by patient sex as indicated in the Sex column. Incidence refers to the percentage of patients in the test set who reach the endpoint by the specified τ . The Endpoint Event column specifies all ICD-10 codes used to identify the disease indicated in the Task column. Category

Task

Sex

Incid.

Endpoint Event

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory

Acute Myocardial Infarction Aortic Aneurysm Aortic Stenosis

1y 5y 5y

153,244 144,435 139,345

0.72% 5.64% 6.05%

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory

Atrial Fibrillation Chronic Ischemic Heart Disease Heart Failure Hypertension Intracranial Hemorrhage Peripheral Artery Disease

5y 5y 3y 3y 1y 5y

145,377 146,743 145,090 155,882 198,214 135,994

23.53% 22.98% 7.73% 23.86% 0.24% 7.88%

Cardiovascular & Circulatory Cardiovascular & Circulatory Death Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Hematology & Sensory

Stroke Venous Thromboembolism All-Cause Mortality Addison’s Disease Benign Prostatic Hyperplasia Cushing Syndrome Dyslipidemia Endometriosis Gout Hyperprolactinemia Hyperthyroidism Hypothyroidism Polycystic Ovary Syndrome Primary Hyperparathyroidism Type 2 Diabetes Acute Pancreatitis Celiac Disease Chronic Kidney Disease Cirrhosis Crohn’s Disease Hepatitis B Hepatitis C Inflammatory Bowel Disease Nephrotic Syndrome Non-Alcoholic Fatty Liver Disease Pyelonephritis Ulcerative Colitis Aplastic Anemia

Male Female Female -

1y 1y 1y 5y 3y 3y 3y 3y 1y 3y 3y 3y 3y 5y 3y 1y 3y 3y 5y 3y 5y 5y 3y 1y 3y 1y 3y 3y

170,662 193,959 187,308 397,187 59,618 378,981 180,048 89,563 169,841 167,424 206,782 197,018 58,712 162,188 165,958 207,377 211,029 155,327 167,141 206,843 191,722 187,549 209,678 408,374 193,722 220,914 209,056 402,106

1.09% 1.67% 5.34% 0.08% 16.06% 0.08% 27.85% 1.57% 0.75% 0.36% 0.96% 5.71% 2.64% 0.87% 7.75% 0.27% 0.41% 10.25% 2.47% 0.48% 0.29% 0.95% 0.88% 0.02% 3.30% 0.04% 0.59% 0.07%

Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory

Chronic Otitis Media Glaucoma Immune Thrombocytopenia Iron Deficiency Anemia Macular Degeneration Meniere Disease Retinal Detachment Sensorineural Hearing Loss

1y 5y 1y 1y 5y 3y 1y 3y

186,458 167,681 220,401 224,567 129,900 176,928 183,849 194,059

0.69% 10.27% 0.06% 1.60% 13.43% 0.16% 0.10% 3.13%

Hematology & Sensory Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic

Vitamin B12 Deficiency Anemia Ankylosing Spondylitis

1y 5y

186,265 378,567

0.07% 0.25%

I21 I71 I06.0, I06.2, I35.0, I35.2, Q23.0, Q25.3 I48 I25 I50 I10 I60, I61, I62 I70.2, I70.3, I70.4, I70.5, I70.6, I70.7, I73.9 I60, I61, I62, I63 I26, I81, I82 death E27.1 N40 E24 E78 N80 M10, M1A E22.1 E05 E02, E03 E28.2 E21.0 E11 K85 K90.0 N18 K70.3, K71.7, K74 K50 B16, B18.0, B18.1 B17.1, B18.2, B19.2 K50, K51, K52.3 N04 K76.0 N10, N11 K51 D61.0, D61.1, D61.2, D61.3, D61.89, D61.9 H65, H66 H40 D69.3 D50 H35.3 H81.0 H33.0, H33.2 H90.3, H90.4, H90.5, H90.6, H90.7, H90.8, H90.A2, H90.A3 D51 M45

Hidradenitis Suppurativa

159,616

0.32%

L73.2

Hip Fracture

176,710

0.41%

S72

Osteoarthritis

157,404

29.88%

M15, M16, M17, M18, M19

Osteoporosis

145,451

17.58%

M80, M81

Psoriasis

204,148

1.26%

L40

Rheumatoid Arthritis

186,955

1.02%

M05, M06 Continued on next page

Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious

Systemic Lupus Erythematosus

199,197

0.36%

M32

Systemic Sclerosis

347,467

0.24%

M34

Female Female Female Male -

5y 5y 1y 1y 1y 1y 1y 5y 3y 5y 3y 1y 5y 3y 5y 5y 5y 5y 5y 5y 5y 3y 5y 5y 6y 5y 5y 5y 3y 5y 5y 2y 5y 5y 3y 5y

90,557 300,331 193,951 211,812 217,105 212,699 191,473 183,668 360,689 124,183 178,230 198,757 402,789 189,414 193,594 130,389 102,025 138,167 368,776 167,883 84,098 210,899 161,655 150,654 138,644 179,264 144,392 173,400 155,112 260,933 102,657 139,318 53,912 194,864 207,762 143,157

5.04% 0.28% 0.34% 3.77% 0.35% 1.45% 1.50% 0.61% 0.10% 2.63% 3.14% 0.07% 0.09% 0.15% 1.39% 2.09% 11.11% 0.80% 0.11% 1.90% 1.99% 0.29% 1.63% 1.06% 8.43% 2.21% 0.93% 1.97% 4.33% 0.23% 1.15% 0.35% 17.24% 1.35% 6.89% 8.45%

G30 G12.21 F31 F32, F33 G40 F41.1 G43 G35 G70.0 G20, G21 G62.9 F20 C91.0 C92.0 C91, C92, C93, C94, C95 C67 C50 C91.1 C92.1, C92.2 C18, C19, C20 C54.1 C71 C64, C65 C22 C34 C43 C90.0 C82, C83, C84, C85 C44 C10 C56 C25 C61 C73 J45 J44

5y 1y 5y 3y

155,555 224,848 176,644 208,952

3.83% 1.55% 0.67% 8.61%

J84 J18 D86 G47.3

Alzheimer’s Disease Amyotrophic Lateral Sclerosis Bipolar Disorder Depression Epilepsy Generalized Anxiety Disorder Migraine Multiple Sclerosis Myasthenia Gravis Parkinson’s Disease Peripheral Neuropathy Schizophrenia Acute Lymphocytic Leukemia Acute Myeloid Leukemia All Leukemia Bladder Cancer Breast Cancer Chronic Lymphocytic Leukemia Chronic Myeloid Leukemia Colorectal Cancer Endometrial Cancer Glioblastoma Kidney Cancer Liver Cancer Lung Cancer Melanoma Multiple Myeloma Non-Hodgkin Lymphoma Non-Melanoma Skin Cancer Oropharyngeal Cancer Ovarian Cancer Pancreatic Cancer Prostate Cancer Thyroid Cancer Asthma Chronic Obstructive Pulmonary Disease Interstitial Lung Disease Pneumonia Sarcoidosis Sleep Apnea

Extended Data Table 7: Evaluation results of new disease onset prediction tasks. We report timedependent performance metrics computed at τ on the test set for 95 disease onset tasks. τ is the threshold duration. Arrows indicate whether higher (↑) or lower (↓) values are better. ICI, Integrated Calibration Index; MCE, Maximum Calibration Error. 10−4 is the default value of the penalizer (λ) for Cox proportional hazards; if convergence cannot be achieved then λ is increased to 1e-3 (∗ ) or 1e-2 (∗∗ ). Mean and 95% CI reported over 100 bootstraps of the test set. Task

Model

Acute Myocardial Infarction

Age-Sex Last Progress A POLLO

Aortic Aneurysm

Age-Sex Last Progress A POLLO

Aortic Stenosis

Age-Sex Last Progress A POLLO

Atrial Fibrillation

Age-Sex Last Progress A POLLO

Chronic Ischemic Heart Disease

Age-Sex Last Progress A POLLO

Heart Failure

Age-Sex Last Progress A POLLO

Hypertension

Age-Sex Last Progress A POLLO

Intracranial Hemorrhage

Age-Sex Last Progress A POLLO

Peripheral Artery Disease

Age-Sex Last Progress A POLLO

Stroke

Age-Sex Last Progress A POLLO

AUC (↑)

Bal. Acc. (↑)

C-Index (↑)

Brier Score (↓)

ICI (↓)

MCE (↓)

0.675

0.641

0.671

0.006

0.001

0.005

(0.645-0.703)

(0.612-0.663)

(0.642-0.700)

(0.005-0.006)

(0.001-0.002)

(0.003-0.008)

0.715

0.653

0.713

0.006

0.001

0.006

(0.685-0.743)

(0.624-0.682)

(0.684-0.741)

(0.005-0.006)

(0.001-0.002)

(0.003-0.009)

0.817

0.740

0.818

0.006

0.001

0.007

(0.793-0.838)

(0.715-0.766)

(0.794-0.839)

(0.005-0.007)

(0.001-0.002)

(0.004-0.011)

0.727

0.670

0.710

0.018

0.004

0.013

(0.705-0.747)

(0.649-0.696)

(0.688-0.729)

(0.017-0.020)

(0.003-0.006)

(0.006-0.019)

0.656

0.624

0.654

0.017

0.004

0.015

(0.634-0.680)

(0.605-0.644)

(0.633-0.677)

(0.016-0.019)

(0.003-0.005)

(0.009-0.022)

0.769

0.707

0.789

0.018

0.003

0.014

(0.753-0.785)

(0.691-0.723)

(0.771-0.803)

(0.017-0.020)

(0.002-0.004)

(0.005-0.022)

0.777

0.715

0.756

0.019

0.005

0.024

(0.762-0.796)

(0.698-0.735)

(0.742-0.776)

(0.018-0.020)

(0.004-0.007)

(0.013-0.032)

0.657

0.629

0.664

0.019

0.005

0.023

(0.635-0.679)

(0.612-0.653)

(0.645-0.686)

(0.017-0.020)

(0.004-0.006)

(0.015-0.031)

0.792

0.725

0.802

0.019

0.006

0.029

(0.772-0.811)

(0.707-0.746)

(0.785-0.821)

(0.018-0.021)

(0.005-0.007)

(0.019-0.037)

0.755

0.692

0.730

0.074

0.012

0.036

(0.745-0.766)

(0.683-0.702)

(0.721-0.742)

(0.071-0.077)

(0.009-0.015)

(0.023-0.050)

0.669

0.619

0.678

0.072

0.009

0.024

(0.658-0.680)

(0.608-0.630)

(0.668-0.687)

(0.070-0.075)

(0.007-0.012)

(0.013-0.036)

0.771

0.711

0.788

0.073

0.008

0.024

(0.761-0.782)

(0.702-0.720)

(0.781-0.797)

(0.070-0.075)

(0.005-0.010)

(0.016-0.034)

0.693

0.646

0.671

0.076

0.011

0.025

(0.682-0.703)

(0.637-0.655)

(0.663-0.681)

(0.074-0.080)

(0.009-0.015)

(0.018-0.035)

0.676

0.647

0.659

0.074

0.007

0.017

(0.665-0.690)

(0.637-0.656)

(0.648-0.671)

(0.071-0.077)

(0.005-0.010)

(0.010-0.028)

0.746

0.685

0.765

0.074

0.005

0.014

(0.736-0.756)

(0.675-0.692)

(0.756-0.774)

(0.072-0.076)

(0.004-0.007)

(0.007-0.024)

0.771

0.705

0.757

0.039

0.005

0.018

(0.756-0.782)

(0.692-0.716)

(0.742-0.766)

(0.037-0.041)

(0.004-0.007)

(0.010-0.029)

0.768

0.704

0.764

0.037

0.004

0.014

(0.757-0.779)

(0.691-0.716)

(0.753-0.774)

(0.035-0.038)

(0.002-0.006)

(0.005-0.027)

0.882

0.808

0.881

0.035

0.004

0.013

(0.873-0.890)

(0.800-0.817)

(0.873-0.888)

(0.034-0.037)

(0.002-0.006)

(0.006-0.023)

0.668

0.630

0.650

0.116

0.017

0.058

(0.661-0.675)

(0.624-0.637)

(0.643-0.656)

(0.114-0.119)

(0.014-0.020)

(0.045-0.073)

0.644

0.610

0.641

0.110

0.009

0.020

(0.636-0.653)

(0.603-0.616)

(0.634-0.648)

(0.108-0.114)

(0.006-0.012)

(0.013-0.028)

0.771

0.703

0.776

0.108

0.013

0.027

(0.765-0.779)

(0.696-0.709)

(0.770-0.782)

(0.105-0.110)

(0.010-0.016)

(0.022-0.034)

0.711

0.663

0.709

0.002

0.000

0.001

(0.677-0.749)

(0.632-0.692)

(0.676-0.747)

(0.002-0.002)

(0.000-0.001)

(0.001-0.003)

0.738

0.682

0.735

0.002

0.001

0.002

(0.694-0.771)

(0.647-0.717)

(0.692-0.768)

(0.002-0.002)

(0.000-0.001)

(0.001-0.003)

0.813

0.735

0.815

0.002

0.001

0.003

(0.783-0.840)

(0.703-0.767)

(0.784-0.841)

(0.002-0.002)

(0.000-0.001)

(0.001-0.005)

0.714

0.674

0.692

0.025

0.005

0.013

(0.697-0.729)

(0.656-0.688)

(0.675-0.708)

(0.023-0.026)

(0.003-0.006)

(0.007-0.020)

0.662

0.616

0.675

0.023

0.004

0.013

(0.636-0.684)

(0.603-0.633)

(0.651-0.694)

(0.021-0.025)

(0.003-0.006)

(0.007-0.021)

0.779

0.711

0.803

0.025

0.006

0.019

(0.761-0.796)

(0.694-0.727)

(0.786-0.819)

(0.023-0.026)

(0.004-0.008)

(0.011-0.028)

0.716

0.659

0.711

0.009

0.001

0.004

(0.700-0.735)

(0.642-0.677)

(0.696-0.731)

(0.008-0.010)

(0.001-0.002)

(0.002-0.005)

0.717

0.658

0.713

0.008

0.001

0.004

(0.698-0.735)

(0.640-0.673)

(0.694-0.731)

(0.008-0.009)

(0.001-0.002)

(0.002-0.007)

0.806

0.729

0.806

0.009

0.002

0.006

(0.787-0.822)

(0.715-0.741)

(0.787-0.822)

(0.008-0.009)

(0.001-0.002)

(0.003-0.009)

Continued on next page

Venous Thromboembolism

Age-Sex Last Progress A POLLO

All-Cause Mortality

Age-Sex Last Progress A POLLO

Addison’s Disease

Age-Sex Last Progress A POLLO

Benign Prostatic Hyperplasia

Age-Sex Last Progress A POLLO

Cushing Syndrome

Age-Sex Last Progress A POLLO

Dyslipidemia

Age-Sex Last Progress A POLLO

Endometriosis

Age-Sex Last Progress A POLLO

Gout

Age-Sex Last Progress A POLLO

Hyperprolactinemia

Age-Sex Last Progress A POLLO

Hyperthyroidism

Age-Sex Last Progress A POLLO

Hypothyroidism

Age-Sex Last Progress A POLLO

Polycystic Ovary Syndrome

Age-Sex Last Progress

0.629

0.608

0.626

0.014

0.003

0.007

(0.614-0.646)

(0.593-0.622)

(0.611-0.643)

(0.013-0.014)

(0.002-0.003)

(0.005-0.009)

0.745

0.689

0.741

0.013

0.002

0.006

(0.730-0.760)

(0.677-0.703)

(0.726-0.756)

(0.012-0.013)

(0.002-0.003)

(0.004-0.009)

0.800

0.731

0.798

0.013

0.002

0.007

(0.786-0.812)

(0.720-0.743)

(0.784-0.810)

(0.012-0.014)

(0.001-0.002)

(0.004-0.011)

0.724

0.659

0.719

0.042

0.005

0.012

(0.715-0.732)

(0.649-0.666)

(0.710-0.727)

(0.040-0.043)

(0.004-0.007)

(0.008-0.019)

0.859

0.785

0.852

0.037

0.007

0.025

(0.851-0.865)

(0.778-0.793)

(0.845-0.858)

(0.036-0.039)

(0.005-0.008)

(0.014-0.036)

0.920

0.849

0.912

0.034

0.005

0.024

(0.916-0.923)

(0.843-0.856)

(0.909-0.916)

(0.033-0.035)

(0.004-0.006)

(0.014-0.034)

0.535

0.564

0.518

0.000

(0.408-0.648)

(0.497-0.607)

(0.393-0.627)

(0.000-0.000)

(0.000-0.001)

0.634

0.666

0.637

0.000

0.001

(0.536-0.754)

(0.594-0.733)

(0.537-0.761)

(0.000-0.000)

(0.001-0.001)

0.790

0.742

0.798

0.000

0.001

(0.685-0.873)

(0.673-0.808)

(0.691-0.879)

(0.000-0.000)

(0.000-0.001)

0.596

0.575

0.590

0.084

0.014

0.035

(0.583-0.614)

(0.565-0.590)

(0.578-0.605)

(0.080-0.087)

(0.010-0.018)

(0.026-0.047)

0.571

0.546

0.585

0.080

0.010

0.025

(0.555-0.585)

(0.532-0.559)

(0.568-0.598)

(0.076-0.084)

(0.006-0.014)

(0.015-0.040)

0.673

0.627

0.704

0.082

0.009

0.024

(0.658-0.686)

(0.615-0.640)

(0.690-0.716)

(0.079-0.085)

(0.006-0.012)

(0.017-0.034)

0.569

0.577

0.571

0.000

(0.505-0.639)

(0.534-0.631)

(0.507-0.641)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.725

0.662

0.727

0.000

0.001

(0.657-0.792)

(0.613-0.717)

(0.660-0.795)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.803

0.699

0.815

0.000

0.001

(0.743-0.861)

(0.659-0.732)

(0.758-0.873)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.655

0.625

0.640

0.140

0.038

0.094

(0.648-0.662)

(0.619-0.631)

(0.633-0.647)

(0.138-0.142)

(0.034-0.041)

(0.078-0.110)

0.630

0.592

0.626

0.133

0.007

0.018

(0.621-0.638)

(0.584-0.599)

(0.620-0.634)

(0.131-0.136)

(0.004-0.011)

(0.009-0.028)

0.719

0.660

0.729

0.131

0.007

0.016

(0.712-0.725)

(0.654-0.665)

(0.724-0.735)

(0.129-0.133)

(0.005-0.009)

(0.011-0.023)

0.653

0.645

0.621

0.009

0.005

0.013

(0.625-0.675)

(0.622-0.668)

(0.592-0.643)

(0.008-0.010)

(0.004-0.006)

(0.010-0.015)

0.608

0.589

0.605

0.009

0.004

0.015

(0.578-0.646)

(0.559-0.622)

(0.575-0.644)

(0.008-0.010)

(0.003-0.005)

(0.011-0.019)

0.705

0.645

0.716

0.009

0.004

0.018

(0.666-0.741)

(0.614-0.674)

(0.678-0.751)

(0.008-0.010)

(0.003-0.005)

(0.012-0.025)

0.736

0.690

0.732

0.006

0.002

0.004

(0.718-0.758)

(0.673-0.708)

(0.714-0.754)

(0.006-0.007)

(0.001-0.002)

(0.002-0.006)

0.735

0.680

0.734

0.006

0.001

0.003

(0.713-0.757)

(0.658-0.703)

(0.712-0.756)

(0.005-0.006)

(0.001-0.001)

(0.001-0.004)

0.830

0.754

0.834

0.006

0.001

0.003

(0.810-0.846)

(0.734-0.771)

(0.814-0.850)

(0.005-0.007)

(0.001-0.001)

(0.001-0.005)

0.778

0.739

0.760

0.002

0.001

0.003

(0.741-0.810)

(0.711-0.769)

(0.722-0.793)

(0.002-0.002)

(0.001-0.001)

(0.001-0.004)

0.684

0.654

0.677

0.002

0.001

0.006

(0.635-0.740)

(0.614-0.694)

(0.627-0.733)

(0.002-0.002)

(0.001-0.002)

(0.004-0.008)

0.853

0.789

0.857

0.002

0.001

0.006

(0.815-0.890)

(0.751-0.819)

(0.820-0.893)

(0.002-0.002)

(0.001-0.001)

(0.003-0.009)

0.585

0.584

0.596

0.005

0.001

0.003

(0.559-0.617)

(0.563-0.609)

(0.571-0.629)

(0.005-0.005)

(0.001-0.002)

(0.002-0.004)

0.592

0.572

0.605

0.005

0.002

0.003

(0.561-0.629)

(0.549-0.597)

(0.574-0.642)

(0.004-0.005)

(0.001-0.002)

(0.002-0.005)

0.715

0.664

0.734

0.005

0.001

0.007

(0.688-0.741)

(0.639-0.688)

(0.707-0.759)

(0.004-0.005)

(0.001-0.002)

(0.004-0.010)

0.601

0.562

0.607

0.029

0.004

0.009

(0.588-0.613)

(0.553-0.571)

(0.595-0.618)

(0.028-0.031)

(0.003-0.006)

(0.006-0.012)

0.616

0.593

0.629

0.027

0.003

0.007

(0.600-0.634)

(0.580-0.605)

(0.613-0.646)

(0.026-0.028)

(0.002-0.004)

(0.004-0.010)

0.746

0.691

0.770

0.029

0.003

0.009

(0.734-0.759)

(0.681-0.701)

(0.759-0.781)

(0.028-0.030)

(0.002-0.004)

(0.004-0.015)

0.726

0.688

0.014

0.007

0.017

(0.706-0.747)

(0.669-0.712)

(0.665-0.709)

(0.013-0.016)

(0.005-0.008)

(0.012-0.023)

0.613

0.596

0.603

0.014

0.005

0.012

(0.580-0.655)

(0.569-0.625)

(0.570-0.644)

(0.012-0.016)

(0.003-0.006)

(0.008-0.017)

Continued on next page

A POLLO Primary Hyperparathyroidism

Age-Sex Last Progress A POLLO

Type 2 Diabetes

Age-Sex Last Progress A POLLO

Acute Pancreatitis

Age-Sex Last Progress A POLLO

Celiac Disease

Age-Sex Last Progress A POLLO

Chronic Kidney Disease

Age-Sex Last Progress A POLLO

Cirrhosis

Age-Sex Last Progress A POLLO

Crohn’s Disease

Age-Sex Last Progress A POLLO

Hepatitis B

Age-Sex Last Progress A POLLO

Hepatitis C

Age-Sex Last Progress A POLLO

Inflammatory Bowel Disease

Age-Sex Last Progress A POLLO

Nephrotic Syndrome

Age-Sex Last Progress A POLLO

Non-Alcoholic Fatty Liver Disease

Age-Sex Last Progress

0.795

0.724

0.774

0.014

0.004

0.014

(0.766-0.832)

(0.700-0.750)

(0.744-0.812)

(0.012-0.016)

(0.003-0.006)

(0.007-0.025)

0.682

0.630

0.679

0.003

0.001

0.003

(0.630-0.732)

(0.590-0.675)

(0.629-0.730)

(0.002-0.003)

(0.001-0.001)

(0.001-0.006)

0.610

0.591

0.621

0.003

0.002

0.007

(0.565-0.658)

(0.554-0.635)

(0.575-0.669)

(0.002-0.003)

(0.001-0.002)

(0.004-0.008)

0.778

0.704

0.825

0.003

0.001

0.004

(0.737-0.819)

(0.657-0.740)

(0.790-0.858)

(0.003-0.004)

(0.001-0.002)

(0.003-0.006)

0.609

0.577

0.599

0.042

0.005

0.012

(0.599-0.620)

(0.567-0.586)

(0.589-0.609)

(0.041-0.044)

(0.003-0.007)

(0.010-0.015)

0.627

0.592

0.625

0.040

0.004

0.010

(0.614-0.639)

(0.582-0.602)

(0.613-0.638)

(0.038-0.042)

(0.003-0.006)

(0.006-0.016)

0.852

0.772

0.854

0.037

0.006

0.020

(0.842-0.860)

(0.763-0.781)

(0.845-0.861)

(0.036-0.039)

(0.005-0.008)

(0.011-0.032)

0.593

0.595

0.591

0.002

0.001

0.002

(0.558-0.624)

(0.563-0.625)

(0.556-0.621)

(0.002-0.003)

(0.000-0.001)

(0.001-0.003)

0.651

0.613

0.648

0.002

0.001

0.002

(0.606-0.694)

(0.578-0.645)

(0.603-0.691)

(0.002-0.002)

(0.000-0.001)

(0.001-0.003)

0.725

0.658

0.728

0.002

0.001

0.002

(0.686-0.759)

(0.627-0.692)

(0.689-0.761)

(0.002-0.002)

(0.000-0.001)

(0.001-0.003)

0.618

0.562

0.614

0.002

0.001

0.002

(0.578-0.655)

(0.524-0.599)

(0.574-0.651)

(0.002-0.003)

(0.001-0.001)

(0.001-0.003)

0.633

0.597

0.638

0.002

0.001

0.003

(0.583-0.685)

(0.558-0.631)

(0.586-0.689)

(0.002-0.003)

(0.001-0.002)

(0.002-0.005)

0.678

0.601

0.704

0.002

0.006

(0.631-0.719)

(0.562-0.639)

(0.656-0.745)

(0.002-0.003)

(0.001-0.002)

(0.004-0.008)

0.749

0.685

0.735

0.053

0.008

0.023

(0.740-0.761)

(0.675-0.695)

(0.726-0.746)

(0.051-0.055)

(0.006-0.009)

(0.014-0.033)

0.711

0.660

0.708

0.053

0.007

0.015

(0.700-0.723)

(0.650-0.671)

(0.697-0.719)

(0.051-0.055)

(0.005-0.009)

(0.010-0.021)

0.863

0.791

0.861

0.046

0.006

0.020

(0.855-0.871)

(0.781-0.799)

(0.853-0.868)

(0.045-0.048)

(0.004-0.008)

(0.010-0.033)

0.634

0.607

0.617

0.008

0.003

0.007

(0.606-0.661)

(0.584-0.635)

(0.591-0.642)

(0.007-0.009)

(0.002-0.004)

(0.004-0.010)

0.680

0.639

0.684

0.007

0.004

0.018

(0.647-0.711)

(0.614-0.665)

(0.650-0.716)

(0.006-0.008)

(0.003-0.004)

(0.014-0.022)

0.865

0.795

0.883

0.008

0.002

0.006

(0.842-0.885)

(0.775-0.817)

(0.863-0.902)

(0.007-0.009)

(0.001-0.002)

(0.003-0.010)

0.583

0.567

0.571

0.002

0.001

(0.539-0.626)

(0.522-0.607)

(0.529-0.612)

(0.002-0.003)

(0.000-0.001)

(0.001-0.002)

0.711

0.656

0.712

0.002

0.001

0.003

(0.654-0.754)

(0.615-0.684)

(0.655-0.756)

(0.002-0.003)

(0.001-0.002)

(0.002-0.004)

0.811

0.733

0.814

0.003

0.001

0.003

(0.769-0.845)

(0.702-0.768)

(0.773-0.848)

(0.002-0.003)

(0.001-0.001)

(0.002-0.006)

0.625

0.596

0.600

0.001

0.000

0.001

(0.504-0.725)

(0.523-0.667)

(0.480-0.701)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.605

0.576

0.629

0.001

0.002

(0.517-0.682)

(0.516-0.634)

(0.541-0.705)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.727

0.653

0.742

0.001

0.002

(0.659-0.789)

(0.582-0.730)

(0.677-0.800)

(0.001-0.002)

(0.000-0.001)

(0.001-0.003)

0.662

0.632

0.634

0.002

0.001

0.002

(0.607-0.707)

(0.594-0.668)

(0.580-0.679)

(0.002-0.003)

(0.001-0.001)

(0.001-0.004)

0.683

0.665

0.694

0.002

0.007

(0.595-0.749)

(0.606-0.703)

(0.608-0.760)

(0.002-0.003)

(0.002-0.002)

(0.006-0.010)

0.888

0.815

0.910

0.002

0.001

0.003

(0.858-0.919)

(0.783-0.850)

(0.880-0.935)

(0.002-0.003)

(0.000-0.001)

(0.001-0.007)

0.524

0.523

0.518

0.005

0.001

0.002

(0.494-0.548)

(0.503-0.550)

(0.489-0.542)

(0.004-0.005)

(0.001-0.001)

(0.002-0.003)

0.621

0.609

0.618

0.004

0.002

0.006

(0.582-0.659)

(0.586-0.637)

(0.579-0.655)

(0.004-0.005)

(0.002-0.002)

(0.004-0.008)

0.757

0.704

0.767

0.004

0.002

0.008

(0.718-0.795)

(0.679-0.729)

(0.729-0.804)

(0.004-0.005)

(0.001-0.002)

(0.005-0.011)

0.625

0.583

0.622

0.000

(0.550-0.692)

(0.512-0.646)

(0.548-0.689)

(0.000-0.000)

(0.000-0.001)

0.739

0.658

0.737

0.000

(0.648-0.821)

(0.599-0.726)

(0.644-0.818)

(0.000-0.000)

(0.000-0.001)

0.803

0.746

0.810

0.000

(0.734-0.866)

(0.694-0.806)

(0.743-0.871)

(0.000-0.000)

(0.000-0.001)

0.521

0.534

0.508

0.018

0.006

0.013

(0.501-0.539)

(0.520-0.546)

(0.487-0.525)

(0.017-0.019)

(0.005-0.007)

(0.010-0.018)

0.597

0.576

0.606

0.018

0.003

0.009

(0.578-0.616)

(0.559-0.591)

(0.586-0.625)

(0.016-0.019)

(0.002-0.004)

(0.005-0.013)

Continued on next page

A POLLO Pyelonephritis

Age-Sex Last Progress A POLLO

Ulcerative Colitis

Age-Sex Last Progress A POLLO

Aplastic Anemia

Age-Sex Last Progress A POLLO

Chronic Otitis Media

Age-Sex Last Progress A POLLO

Glaucoma

Age-Sex Last Progress A POLLO

Immune Thrombocytopenia

Age-Sex Last Progress A POLLO

Iron Deficiency Anemia

Age-Sex Last Progress A POLLO

Macular Degeneration

Age-Sex Last Progress A POLLO

Meniere Disease

Age-Sex Last Progress A POLLO

Retinal Detachment

Age-Sex Last Progress A POLLO

Sensorineural Hearing Loss

Age-Sex Last Progress A POLLO

Vitamin B12 Deficiency Anemia

Age-Sex Last Progress

0.771

0.701

0.785

0.018

0.002

0.009

(0.755-0.785)

(0.685-0.716)

(0.770-0.799)

(0.017-0.018)

(0.001-0.003)

(0.004-0.014)

0.573

0.539

0.575

0.000

0.001

(0.456-0.683)

(0.456-0.634)

(0.458-0.685)

(0.000-0.000)

(0.000-0.001)

0.567

0.578

0.568

0.000

0.001

(0.440-0.674)

(0.466-0.660)

(0.440-0.675)

(0.000-0.000)

(0.000-0.001)

0.733

0.671

0.737

0.000

0.002

(0.649-0.811)

(0.608-0.742)

(0.655-0.812)

(0.000-0.001)

(0.001-0.004)

0.506

0.523

0.512

0.003

0.001

0.002

(0.469-0.539)

(0.490-0.553)

(0.475-0.544)

(0.003-0.004)

(0.001-0.001)

(0.001-0.004)

0.629

0.623

0.622

0.003

0.001

0.005

(0.580-0.680)

(0.585-0.656)

(0.573-0.673)

(0.003-0.004)

(0.001-0.002)

(0.003-0.007)

0.681

0.659

0.695

0.003

0.002

0.005

(0.636-0.722)

(0.629-0.693)

(0.650-0.734)

(0.003-0.004)

(0.001-0.002)

(0.003-0.007)

0.700

0.643

0.693

0.000

(0.642-0.760)

(0.589-0.693)

(0.636-0.752)

(0.000-0.000)

0.706

0.645

0.714

0.000

0.001

(0.620-0.769)

(0.578-0.706)

(0.628-0.776)

(0.000-0.000)

(0.000-0.001)

0.836

0.784

0.849

0.000

0.001

(0.785-0.893)

(0.740-0.819)

(0.801-0.905)

(0.000-0.000)

(0.000-0.001)

0.642

0.618

0.636

0.006

0.002

0.007

(0.619-0.665)

(0.600-0.636)

(0.613-0.658)

(0.005-0.006)

(0.002-0.003)

(0.004-0.010)

0.675

0.626

0.674

0.006

0.001

0.003

(0.648-0.702)

(0.605-0.651)

(0.647-0.700)

(0.005-0.006)

(0.001-0.001)

(0.001-0.004)

0.765

0.690

0.769

0.006

0.001

0.003

(0.745-0.787)

(0.667-0.709)

(0.750-0.792)

(0.005-0.006)

(0.001-0.001)

(0.001-0.005)

0.643

0.611

0.626

0.035

0.006

0.016

(0.626-0.658)

(0.597-0.624)

(0.610-0.640)

(0.033-0.037)

(0.004-0.008)

(0.009-0.024)

0.572

0.541

0.602

0.034

0.004

0.011

(0.553-0.589)

(0.526-0.553)

(0.584-0.618)

(0.032-0.036)

(0.002-0.006)

(0.006-0.017)

0.622

0.592

0.691

0.036

0.004

0.013

(0.605-0.641)

(0.578-0.607)

(0.673-0.708)

(0.034-0.038)

(0.003-0.006)

(0.007-0.020)

0.597

0.548

0.595

0.000

0.001

(0.518-0.669)

(0.474-0.616)

(0.517-0.666)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.627

0.604

0.625

0.000

0.001

(0.508-0.704)

(0.528-0.667)

(0.506-0.702)

(0.000-0.001)

(0.001-0.002)

0.735

0.680

0.737

0.001

0.000

0.001

(0.637-0.816)

(0.599-0.745)

(0.639-0.817)

(0.000-0.001)

(0.001-0.002)

0.575

0.578

0.579

0.013

0.004

0.010

(0.562-0.587)

(0.568-0.586)

(0.567-0.592)

(0.012-0.014)

(0.003-0.005)

(0.007-0.014)

0.661

0.624

0.658

0.013

0.002

0.004

(0.644-0.678)

(0.611-0.641)

(0.641-0.675)

(0.012-0.014)

(0.001-0.002)

(0.002-0.007)

0.766

0.699

0.768

0.013

0.002

0.005

(0.749-0.779)

(0.683-0.712)

(0.752-0.782)

(0.012-0.013)

(0.001-0.002)

(0.002-0.009)

0.712

0.657

0.692

0.044

0.008

0.017

(0.699-0.727)

(0.644-0.668)

(0.680-0.705)

(0.042-0.046)

(0.006-0.010)

(0.012-0.026)

0.594

0.576

0.604

0.045

0.007

0.022

(0.575-0.615)

(0.562-0.590)

(0.585-0.623)

(0.042-0.048)

(0.005-0.009)

(0.015-0.031)

0.714

0.669

0.752

0.045

0.004

0.011

(0.699-0.731)

(0.657-0.681)

(0.738-0.766)

(0.043-0.047)

(0.003-0.006)

(0.006-0.019)

0.519

0.554

0.511

0.001

0.000

0.001

(0.466-0.590)

(0.496-0.604)

(0.458-0.580)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.598

0.527

0.601

0.001

0.002

(0.519-0.690)

(0.482-0.571)

(0.522-0.693)

(0.001-0.001)

(0.002-0.003)

0.629

0.604

0.649

0.001

0.003

(0.548-0.701)

(0.540-0.656)

(0.569-0.722)

(0.001-0.001)

(0.002-0.004)

0.585

0.539

0.582

0.001

0.000

0.001

(0.523-0.649)

(0.487-0.594)

(0.521-0.647)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.577

0.571

0.578

0.001

0.003

(0.497-0.664)

(0.515-0.628)

(0.499-0.665)

(0.001-0.001)

(0.002-0.004)

0.680

0.627

0.691

0.001

0.004

(0.598-0.757)

(0.565-0.696)

(0.610-0.767)

(0.001-0.001)

(0.003-0.005)

0.719

0.663

0.709

0.017

0.002

0.006

(0.703-0.731)

(0.648-0.676)

(0.692-0.721)

(0.016-0.017)

(0.001-0.003)

(0.003-0.011)

0.638

0.610

0.645

0.016

0.002

0.007

(0.622-0.652)

(0.597-0.621)

(0.630-0.659)

(0.015-0.017)

(0.002-0.003)

(0.004-0.011)

0.745

0.679

0.769

0.017

0.002

0.007

(0.729-0.756)

(0.667-0.691)

(0.754-0.778)

(0.016-0.017)

(0.002-0.003)

(0.003-0.011)

0.588

0.565

0.591

0.001

0.000

0.001

(0.523-0.653)

(0.505-0.627)

(0.527-0.656)

(0.000-0.001)

(0.001-0.001)

0.450

0.440

0.449

0.001

0.003

(0.374-0.559)

(0.379-0.512)

(0.373-0.557)

(0.000-0.001)

(0.001-0.001)

(0.002-0.004)

Continued on next page

A POLLO Ankylosing Spondylitis

Age-Sex Last Progress A POLLO

Hidradenitis Suppurativa

Age-Sex Last Progress A POLLO

Hip Fracture

Age-Sex Last Progress A POLLO

Osteoarthritis

Age-Sex Last Progress A POLLO

Osteoporosis

Age-Sex Last Progress A POLLO

Psoriasis

Age-Sex Last Progress A POLLO

Rheumatoid Arthritis

Age-Sex Last Progress A POLLO

Systemic Lupus Erythematosus

Age-Sex Last Progress A POLLO

Systemic Sclerosis

Age-Sex Last Progress A POLLO

Alzheimer’s Disease

Age-Sex Last Progress A POLLO

Amyotrophic Lateral Sclerosis

Age-Sex Last Progress A POLLO

Bipolar Disorder

Age-Sex Last Progress

0.680

0.584

0.685

0.001

0.005

(0.609-0.750)

(0.521-0.646)

(0.614-0.754)

(0.001-0.001)

(0.003-0.005)

0.601

0.588

0.580

0.001

0.000

0.001

(0.548-0.649)

(0.548-0.625)

(0.526-0.629)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.616

0.569

0.626

0.001

0.000

0.001

(0.542-0.681)

(0.515-0.620)

(0.553-0.689)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.775

0.715

0.784

0.001

0.004

(0.700-0.835)

(0.670-0.754)

(0.710-0.842)

(0.001-0.001)

(0.000-0.001)

(0.002-0.005)

0.691

0.634

0.680

0.002

0.001

0.005

(0.649-0.738)

(0.587-0.689)

(0.638-0.727)

(0.001-0.002)

(0.001-0.001)

(0.004-0.007)

0.571

0.550

0.577

0.002

0.009

(0.493-0.624)

(0.497-0.590)

(0.498-0.632)

(0.002-0.002)

(0.001-0.002)

(0.007-0.010)

0.750

0.689

0.759

0.002

0.011

(0.706-0.799)

(0.642-0.741)

(0.715-0.806)

(0.002-0.003)

(0.002-0.002)

(0.009-0.013)

0.825

0.757

0.823

0.003

0.001

0.002

(0.799-0.853)

(0.731-0.787)

(0.796-0.850)

(0.003-0.004)

(0.000-0.001)

(0.001-0.004)

0.773

0.714

0.770

0.003

0.001

0.003

(0.746-0.804)

(0.688-0.740)

(0.743-0.800)

(0.003-0.004)

(0.001-0.001)

(0.001-0.005)

0.867

0.785

0.866

0.003

0.001

0.005

(0.848-0.883)

(0.761-0.809)

(0.848-0.882)

(0.003-0.004)

(0.001-0.001)

(0.003-0.008)

0.616

0.594

0.605

0.104

0.022

0.057

(0.606-0.626)

(0.586-0.604)

(0.597-0.612)

(0.101-0.107)

(0.019-0.026)

(0.041-0.070)

0.605

0.581

0.623

0.102

0.009

0.026

(0.592-0.617)

(0.570-0.590)

(0.611-0.633)

(0.100-0.105)

(0.007-0.012)

(0.016-0.036)

0.681

0.637

0.724

0.104

0.007

0.016

(0.673-0.694)

(0.629-0.646)

(0.717-0.735)

(0.101-0.106)

(0.004-0.009)

(0.010-0.025)

0.737

0.706

0.726

0.058

0.019

0.048

(0.728-0.746)

(0.697-0.717)

(0.718-0.735)

(0.055-0.060)

(0.017-0.022)

(0.037-0.059)

0.648

0.620

0.658

0.057

0.009

0.020

(0.634-0.661)

(0.608-0.631)

(0.646-0.669)

(0.054-0.059)

(0.006-0.011)

(0.013-0.029)

0.768

0.709

0.781

0.057

0.009

0.028

(0.757-0.779)

(0.698-0.718)

(0.773-0.791)

(0.055-0.059)

(0.006-0.011)

(0.018-0.042)

0.539

0.536

0.539

0.007

0.002

0.003

(0.517-0.561)

(0.521-0.551)

(0.517-0.562)

(0.006-0.008)

(0.001-0.002)

(0.002-0.005)

0.583

0.553

0.592

0.007

0.002

0.004

(0.553-0.614)

(0.526-0.586)

(0.562-0.623)

(0.006-0.007)

(0.001-0.003)

(0.003-0.006)

0.647

0.610

0.679

0.007

0.002

0.005

(0.617-0.672)

(0.586-0.632)

(0.651-0.704)

(0.006-0.008)

(0.001-0.003)

(0.003-0.006)

0.602

0.582

0.607

0.005

0.001

0.003

(0.568-0.627)

(0.548-0.604)

(0.573-0.632)

(0.005-0.006)

(0.001-0.002)

(0.002-0.005)

0.627

0.591

0.635

0.005

0.002

0.007

(0.585-0.669)

(0.569-0.622)

(0.593-0.678)

(0.005-0.006)

(0.002-0.003)

(0.005-0.010)

0.750

0.695

0.778

0.005

0.002

0.007

(0.723-0.779)

(0.668-0.718)

(0.751-0.806)

(0.005-0.006)

(0.001-0.002)

(0.004-0.010)

0.699

0.644

0.697

0.001

0.002

(0.637-0.753)

(0.595-0.694)

(0.639-0.751)

(0.001-0.001)

(0.001-0.003)

0.680

0.599

0.681

0.001

0.002

(0.608-0.760)

(0.544-0.661)

(0.609-0.763)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.762

0.686

0.783

0.001

0.005

(0.697-0.828)

(0.613-0.743)

(0.718-0.845)

(0.001-0.001)

(0.003-0.007)

0.679

0.651

0.683

0.001

0.000

0.001

(0.616-0.741)

(0.615-0.693)

(0.622-0.744)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.681

0.675

0.700

0.001

0.000

0.001

(0.620-0.744)

(0.623-0.718)

(0.640-0.762)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.775

0.759

0.797

0.001

0.000

0.001

(0.708-0.829)

(0.719-0.792)

(0.733-0.852)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.752

0.717

0.739

0.017

0.005

0.017

(0.723-0.779)

(0.696-0.738)

(0.713-0.765)

(0.015-0.019)

(0.003-0.007)

(0.009-0.028)

0.672

0.635

0.674

0.017

0.004

0.011

(0.638-0.708)

(0.608-0.662)

(0.640-0.711)

(0.015-0.019)

(0.003-0.006)

(0.007-0.018)

0.752

0.705

0.780

0.017

0.006

0.017

(0.720-0.785)

(0.681-0.731)

(0.749-0.813)

(0.015-0.019)

(0.004-0.008)

(0.010-0.024)

0.615

0.567

0.599

0.001

0.000

0.001

(0.565-0.660)

(0.518-0.610)

(0.548-0.644)

(0.001-0.001)

(0.000-0.001)

(0.000-0.002)

0.662

0.733

0.669

0.001

0.000

0.001

(0.580-0.764)

(0.690-0.784)

(0.587-0.770)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.784

0.724

0.800

0.001

0.000

0.001

(0.720-0.837)

(0.678-0.771)

(0.736-0.850)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.631

0.604

0.624

0.003

0.001

0.003

(0.601-0.659)

(0.578-0.630)

(0.595-0.653)

(0.003-0.003)

(0.001-0.001)

(0.002-0.004)

0.722

0.674

0.721

0.003

0.001

0.005

(0.685-0.757)

(0.640-0.704)

(0.684-0.756)

(0.002-0.003)

(0.001-0.001)

(0.002-0.007)

Continued on next page

A POLLO Depression

Age-Sex Last Progress A POLLO

Epilepsy

Age-Sex Last Progress A POLLO

Generalized Anxiety Disorder

Age-Sex Last Progress A POLLO

Migraine

Age-Sex Last Progress A POLLO

Multiple Sclerosis

Age-Sex Last Progress A POLLO

Myasthenia Gravis

Age-Sex Last Progress A POLLO

Parkinson’s Disease

Age-Sex Last Progress A POLLO

Peripheral Neuropathy

Age-Sex Last Progress A POLLO

Schizophrenia

Age-Sex Last Progress A POLLO

Acute Lymphocytic Leukemia

Age-Sex Last Progress A POLLO

Acute Myeloid Leukemia

Age-Sex Last Progress A POLLO

All Leukemia

Age-Sex Last Progress

0.865

0.793

0.865

0.003

0.001

0.006

(0.845-0.888)

(0.773-0.819)

(0.845-0.888)

(0.003-0.003)

(0.001-0.002)

(0.004-0.009)

0.547

0.537

0.547

0.031

0.003

0.009

(0.538-0.557)

(0.529-0.543)

(0.537-0.556)

(0.029-0.032)

(0.002-0.004)

(0.006-0.013)

0.630

0.595

0.629

0.029

0.003

0.006

(0.619-0.642)

(0.585-0.605)

(0.618-0.641)

(0.028-0.030)

(0.002-0.003)

(0.004-0.008)

0.721

0.662

0.725

0.030

0.003

0.007

(0.712-0.730)

(0.652-0.670)

(0.716-0.734)

(0.029-0.031)

(0.002-0.004)

(0.004-0.011)

0.587

0.569

0.582

0.003

0.001

0.002

(0.548-0.621)

(0.536-0.599)

(0.542-0.617)

(0.003-0.003)

(0.000-0.001)

(0.001-0.003)

0.736

0.692

0.732

0.003

0.001

0.003

(0.695-0.775)

(0.657-0.722)

(0.691-0.771)

(0.002-0.003)

(0.001-0.001)

(0.001-0.005)

0.861

0.796

0.862

0.003

0.001

0.003

(0.831-0.887)

(0.770-0.820)

(0.833-0.887)

(0.002-0.003)

(0.000-0.001)

(0.001-0.006)

0.587

0.560

0.586

0.012

0.002

0.004

(0.571-0.605)

(0.545-0.576)

(0.570-0.604)

(0.011-0.013)

(0.001-0.002)

(0.002-0.005)

0.624

0.592

0.626

0.012

0.002

0.004

(0.610-0.637)

(0.579-0.608)

(0.612-0.638)

(0.011-0.013)

(0.001-0.002)

(0.002-0.007)

0.699

0.646

0.705

0.012

0.002

0.005

(0.685-0.712)

(0.632-0.660)

(0.691-0.718)

(0.011-0.013)

(0.001-0.003)

(0.003-0.008)

0.687

0.655

0.683

0.012

0.003

0.007

(0.669-0.701)

(0.642-0.669)

(0.666-0.697)

(0.012-0.013)

(0.002-0.004)

(0.005-0.011)

0.683

0.636

0.683

0.012

0.002

0.004

(0.663-0.704)

(0.620-0.653)

(0.663-0.704)

(0.011-0.013)

(0.001-0.002)

(0.002-0.007)

0.743

0.678

0.747

0.012

0.001

0.003

(0.722-0.760)

(0.661-0.694)

(0.726-0.763)

(0.011-0.013)

(0.001-0.002)

(0.002-0.006)

0.677

0.614

0.668

0.002

0.001

0.003

(0.614-0.725)

(0.582-0.650)

(0.605-0.716)

(0.002-0.002)

(0.001-0.001)

(0.001-0.005)

0.672

0.618

0.668

0.002

0.005

(0.598-0.734)

(0.580-0.658)

(0.595-0.733)

(0.001-0.002)

(0.003-0.007)

0.784

0.749

0.796

0.002

0.001

0.002

(0.738-0.842)

(0.698-0.791)

(0.751-0.853)

(0.002-0.003)

(0.001-0.001)

(0.001-0.003)

0.648

0.597

0.640

0.001

0.000

0.001

(0.588-0.701)

(0.556-0.634)

(0.582-0.693)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.692

0.610

0.698

0.001

0.000

0.001

(0.634-0.750)

(0.555-0.660)

(0.639-0.756)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.755

0.680

0.764

0.001

0.000

0.001

(0.719-0.796)

(0.646-0.719)

(0.729-0.804)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.706

0.647

0.692

0.009

0.002

0.007

(0.674-0.747)

(0.618-0.677)

(0.660-0.735)

(0.008-0.010)

(0.001-0.004)

(0.003-0.014)

0.636

0.596

0.654

0.009

0.005

0.024

(0.589-0.675)

(0.566-0.631)

(0.610-0.691)

(0.008-0.010)

(0.004-0.006)

(0.019-0.029)

0.736

0.688

0.756

0.010

0.006

0.028

(0.691-0.774)

(0.649-0.720)

(0.716-0.794)

(0.009-0.011)

(0.005-0.007)

(0.021-0.035)

0.635

0.606

0.626

0.017

0.003

0.006

(0.621-0.650)

(0.593-0.621)

(0.612-0.640)

(0.016-0.018)

(0.002-0.004)

(0.004-0.010)

0.645

0.609

0.648

0.016

0.003

0.009

(0.630-0.664)

(0.592-0.625)

(0.633-0.667)

(0.015-0.017)

(0.002-0.004)

(0.005-0.014)

0.734

0.674

0.756

0.017

0.002

0.005

(0.721-0.746)

(0.664-0.685)

(0.744-0.768)

(0.016-0.017)

(0.001-0.003)

(0.002-0.009)

0.654

0.624

0.648

0.001

0.000

0.001

(0.575-0.755)

(0.554-0.704)

(0.570-0.749)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.840

0.741

0.839

0.001

0.000

0.002

(0.770-0.912)

(0.674-0.826)

(0.769-0.911)

(0.000-0.001)

(0.001-0.003)

0.920

0.818

0.919

0.001

0.000

0.001

(0.892-0.952)

(0.755-0.900)

(0.891-0.952)

(0.000-0.001)

(0.000-0.000)

(0.000-0.002)

0.619

0.518

0.603

0.000

(0.543-0.685)

(0.487-0.556)

(0.528-0.669)

(0.000-0.000)

(0.000-0.001)

0.621

0.641

0.633

0.000

0.001

(0.483-0.713)

(0.574-0.701)

(0.497-0.725)

(0.000-0.000)

(0.000-0.001)

0.764

0.757

0.797

0.000

0.001

(0.658-0.872)

(0.690-0.823)

(0.694-0.894)

(0.000-0.000)

(0.000-0.002)

0.700

0.646

0.691

0.001

0.000

0.001

(0.627-0.770)

(0.575-0.698)

(0.617-0.759)

(0.001-0.001)

(0.000-0.001)

(0.000-0.002)

0.760

0.719

0.763

0.001

0.000

0.002

(0.673-0.838)

(0.650-0.797)

(0.675-0.840)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.839

0.774

0.850

0.001

0.000

0.002

(0.781-0.900)

(0.727-0.830)

(0.790-0.908)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.718

0.648

0.701

0.004

0.001

0.004

(0.682-0.753)

(0.614-0.684)

(0.666-0.736)

(0.004-0.005)

(0.001-0.002)

(0.002-0.006)

0.654

0.636

0.659

0.004

0.002

0.009

(0.615-0.699)

(0.608-0.670)

(0.620-0.704)

(0.004-0.005)

(0.002-0.003)

(0.006-0.011)

Continued on next page

A POLLO Bladder Cancer

Age-Sex Last Progress A POLLO

Breast Cancer

Age-Sex Last Progress A POLLO

Chronic Lymphocytic Leukemia

Age-Sex Last Progress A POLLO

Chronic Myeloid Leukemia

Age-Sex Last Progress A POLLO

Colorectal Cancer

Age-Sex Last Progress A POLLO

Endometrial Cancer

Age-Sex Last Progress A POLLO

Glioblastoma

Age-Sex Last Progress A POLLO

Kidney Cancer

Age-Sex Last Progress A POLLO

Liver Cancer

Age-Sex Last Progress A POLLO

Lung Cancer

Age-Sex Last Progress A POLLO

Melanoma

Age-Sex Last Progress A POLLO

Multiple Myeloma

Age-Sex Last Progress

0.726

0.682

0.744

0.005

0.002

0.007

(0.681-0.759)

(0.648-0.708)

(0.701-0.777)

(0.004-0.005)

(0.002-0.003)

(0.004-0.010)

0.776

0.713

0.757

0.006

0.002

0.005

(0.745-0.802)

(0.682-0.740)

(0.725-0.782)

(0.005-0.007)

(0.001-0.002)

(0.003-0.009)

0.731

0.646

0.725

0.006

0.002

0.010

(0.692-0.764)

(0.611-0.689)

(0.688-0.759)

(0.005-0.006)

(0.002-0.003)

(0.006-0.014)

0.757

0.712

0.783

0.006

0.002

0.009

(0.715-0.795)

(0.685-0.741)

(0.741-0.819)

(0.005-0.007)

(0.002-0.003)

(0.005-0.013)

0.596

0.587

0.586

0.037

0.009

0.020

(0.579-0.617)

(0.569-0.601)

(0.570-0.604)

(0.035-0.040)

(0.007-0.011)

(0.014-0.026)

0.574

0.551

0.587

0.033

0.007

0.018

(0.556-0.599)

(0.534-0.564)

(0.568-0.612)

(0.031-0.035)

(0.005-0.010)

(0.011-0.027)

0.674

0.648

0.712

0.035

0.007

0.015

(0.658-0.689)

(0.634-0.664)

(0.698-0.726)

(0.032-0.037)

(0.005-0.009)

(0.010-0.021)

0.704

0.630

0.685

0.002

0.001

0.003

(0.657-0.749)

(0.591-0.673)

(0.637-0.734)

(0.002-0.003)

(0.001-0.002)

(0.001-0.005)

0.533

0.501

0.556

0.002

0.003

0.020

(0.459-0.601)

(0.466-0.543)

(0.478-0.623)

(0.002-0.003)

(0.003-0.004)

(0.018-0.023)

0.722

0.670

0.721

0.002

0.003

0.016

(0.662-0.779)

(0.631-0.713)

(0.661-0.777)

(0.002-0.003)

(0.013-0.019)

0.682

0.617

0.670

0.000

(0.614-0.728)

(0.535-0.674)

(0.600-0.716)

(0.000-0.000)

(0.000-0.001)

0.705

0.648

0.717

0.000

0.001

(0.636-0.755)

(0.577-0.714)

(0.649-0.765)

(0.000-0.001)

(0.000-0.000)

(0.000-0.001)

0.643

0.616

0.668

0.000

0.002

(0.562-0.736)

(0.570-0.673)

(0.585-0.760)

(0.000-0.001)

(0.001-0.002)

0.665

0.634

0.647

0.006

0.001

0.003

(0.620-0.706)

(0.604-0.666)

(0.601-0.688)

(0.005-0.006)

(0.001-0.002)

(0.002-0.007)

0.614

0.580

0.619

0.005

0.002

0.009

(0.568-0.656)

(0.551-0.602)

(0.574-0.661)

(0.004-0.006)

(0.002-0.003)

(0.006-0.012)

0.645

0.609

0.668

0.005

0.003

0.012

(0.607-0.693)

(0.575-0.649)

(0.631-0.715)

(0.005-0.006)

(0.002-0.004)

(0.009-0.016)

0.612

0.573

0.602

0.006

0.002

0.005

(0.571-0.652)

(0.534-0.610)

(0.563-0.642)

(0.005-0.007)

(0.002-0.003)

(0.004-0.007)

0.483

0.509

0.518

0.006

0.005

0.023

(0.415-0.548)

(0.479-0.550)

(0.450-0.580)

(0.005-0.007)

(0.004-0.006)

(0.019-0.027)

0.685

0.638

0.711

0.006

0.005

0.023

(0.640-0.743)

(0.603-0.675)

(0.668-0.767)

(0.005-0.007)

(0.003-0.006)

(0.018-0.028)

0.579

0.570

0.001

0.002

(0.524-0.633)

(0.530-0.611)

(0.516-0.625)

(0.001-0.002)

(0.000-0.001)

(0.001-0.003)

0.635

0.586

0.633

0.001

0.007

(0.569-0.692)

(0.538-0.637)

(0.567-0.690)

(0.001-0.002)

(0.005-0.008)

0.810

0.763

0.817

0.001

0.002

(0.765-0.859)

(0.720-0.800)

(0.771-0.865)

(0.001-0.002)

(0.000-0.001)

(0.001-0.004)

0.711

0.661

0.693

0.005

0.001

0.004

(0.684-0.744)

(0.635-0.691)

(0.666-0.725)

(0.004-0.005)

(0.001-0.002)

(0.002-0.007)

0.650

0.616

0.666

0.004

0.002

0.006

(0.588-0.698)

(0.568-0.649)

(0.605-0.713)

(0.003-0.005)

(0.002-0.003)

(0.004-0.008)

0.779

0.700

0.802

0.004

0.001

0.004

(0.729-0.820)

(0.663-0.737)

(0.756-0.842)

(0.004-0.005)

(0.001-0.002)

(0.002-0.006)

0.714

0.657

0.695

0.003

0.001

0.003

(0.663-0.768)

(0.619-0.702)

(0.641-0.749)

(0.003-0.004)

(0.001-0.002)

(0.002-0.007)

0.727

0.694

0.729

0.003

0.001

0.005

(0.663-0.792)

(0.640-0.739)

(0.665-0.794)

(0.002-0.003)

(0.001-0.002)

(0.002-0.007)

0.853

0.768

0.862

0.003

0.001

0.004

(0.807-0.895)

(0.722-0.812)

(0.819-0.903)

(0.003-0.004)

(0.001-0.002)

(0.002-0.007)

0.672

0.624

0.647

0.017

0.006

0.019

(0.653-0.693)

(0.610-0.638)

(0.628-0.666)

(0.015-0.018)

(0.005-0.007)

(0.014-0.024)

0.639

0.627

0.649

0.015

0.005

0.018

(0.610-0.670)

(0.605-0.646)

(0.622-0.680)

(0.014-0.017)

(0.003-0.006)

(0.010-0.024)

0.754

0.707

0.789

0.016

0.006

0.027

(0.729-0.778)

(0.688-0.723)

(0.767-0.814)

(0.015-0.018)

(0.005-0.007)

(0.016-0.036)

0.663

0.640

0.651

0.007

0.002

0.004

(0.632-0.697)

(0.617-0.664)

(0.621-0.683)

(0.006-0.008)

(0.001-0.002)

(0.002-0.006)

0.592

0.566

0.602

0.006

0.002

0.006

(0.555-0.632)

(0.538-0.593)

(0.565-0.642)

(0.006-0.007)

(0.002-0.003)

(0.004-0.009)

0.720

0.663

0.753

0.007

0.002

0.006

(0.689-0.753)

(0.634-0.688)

(0.725-0.784)

(0.006-0.008)

(0.001-0.003)

(0.003-0.010)

0.649

0.591

0.637

0.003

0.001

0.004

(0.593-0.698)

(0.551-0.632)

(0.582-0.688)

(0.002-0.004)

(0.001-0.002)

(0.002-0.007)

0.549

0.547

0.538

0.003

0.006

(0.487-0.615)

(0.499-0.593)

(0.475-0.604)

(0.003-0.004)

(0.002-0.004)

(0.004-0.008)

Continued on next page

A POLLO Non-Hodgkin Lymphoma

Age-Sex Last Progress A POLLO

Non-Melanoma Skin Cancer

Age-Sex Last Progress A POLLO

Oropharyngeal Cancer

Age-Sex Last Progress A POLLO

Ovarian Cancer

Age-Sex Last Progress A POLLO

Pancreatic Cancer

Age-Sex Last Progress A POLLO

Prostate Cancer

Age-Sex Last Progress A POLLO

Thyroid Cancer

Age-Sex Last Progress A POLLO

Asthma

Age-Sex Last Progress A POLLO

Chronic Obstructive Pulmonary Disease

Age-Sex Last Progress A POLLO

Interstitial Lung Disease

Age-Sex Last Progress A POLLO

Pneumonia

Age-Sex Last Progress A POLLO

Sarcoidosis

Age-Sex

0.683

0.642

0.714

0.003

0.002

0.005

(0.631-0.741)

(0.596-0.688)

(0.665-0.769)

(0.003-0.004)

(0.002-0.003)

(0.003-0.009)

0.675

0.630

0.658

0.006

0.001

0.004

(0.630-0.716)

(0.601-0.653)

(0.617-0.697)

(0.006-0.007)

(0.001-0.002)

(0.002-0.008)

0.560

0.522

0.580

0.006

0.002

0.005

(0.515-0.610)

(0.492-0.552)

(0.534-0.628)

(0.005-0.006)

(0.001-0.003)

(0.003-0.008)

0.721

0.651

0.754

0.006

0.003

0.013

(0.686-0.756)

(0.623-0.682)

(0.719-0.785)

(0.006-0.007)

(0.003-0.004)

(0.009-0.017)

0.662

0.621

0.651

0.024

0.002

0.006

(0.647-0.677)

(0.604-0.633)

(0.636-0.666)

(0.022-0.025)

(0.001-0.004)

(0.003-0.011)

0.602

0.576

0.614

0.021

0.003

0.007

(0.583-0.620)

(0.561-0.592)

(0.596-0.633)

(0.020-0.023)

(0.002-0.004)

(0.004-0.011)

0.737

0.676

0.759

0.023

0.002

0.005

(0.720-0.751)

(0.662-0.689)

(0.743-0.772)

(0.022-0.024)

(0.001-0.003)

(0.002-0.008)

0.720

0.727

0.711

0.001

0.000

0.001

(0.634-0.780)

(0.683-0.767)

(0.625-0.771)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.787

0.722

0.789

0.001

0.000

0.001

(0.707-0.866)

(0.680-0.760)

(0.707-0.869)

(0.001-0.001)

(0.000-0.000)

(0.000-0.002)

0.851

0.797

0.848

0.001

0.000

0.001

(0.786-0.901)

(0.748-0.842)

(0.782-0.899)

(0.001-0.001)

(0.000-0.000)

(0.000-0.001)

0.644

0.612

0.631

0.003

0.002

0.004

(0.602-0.685)

(0.578-0.639)

(0.593-0.672)

(0.003-0.004)

(0.001-0.003)

(0.003-0.007)

0.638

0.608

0.621

0.003

0.006

(0.576-0.701)

(0.560-0.670)

(0.559-0.688)

(0.003-0.004)

(0.002-0.004)

(0.004-0.010)

0.707

0.665

0.741

0.003

0.002

0.009

(0.659-0.757)

(0.624-0.710)

(0.696-0.789)

(0.003-0.004)

(0.002-0.003)

(0.005-0.012)

0.656

0.586

0.651

0.002

0.001

0.002

(0.611-0.696)

(0.536-0.627)

(0.605-0.692)

(0.002-0.003)

(0.000-0.001)

(0.001-0.003)

0.608

0.549

0.608

0.002

0.001

0.004

(0.543-0.663)

(0.500-0.593)

(0.542-0.662)

(0.002-0.002)

(0.001-0.002)

(0.002-0.006)

0.704

0.648

0.708

0.002

0.009

(0.647-0.761)

(0.594-0.692)

(0.650-0.765)

(0.002-0.002)

(0.001-0.002)

(0.006-0.011)

0.606

0.593

0.591

0.051

0.013

0.033

(0.589-0.630)

(0.577-0.614)

(0.572-0.610)

(0.047-0.055)

(0.009-0.016)

(0.019-0.051)

0.611

0.590

0.617

0.047

0.011

0.026

(0.583-0.634)

(0.573-0.606)

(0.595-0.641)

(0.043-0.052)

(0.008-0.015)

(0.017-0.036)

0.671

0.654

0.711

0.051

0.010

0.025

(0.649-0.696)

(0.641-0.671)

(0.690-0.731)

(0.047-0.055)

(0.007-0.013)

(0.015-0.037)

0.481

0.509

0.499

0.004

0.001

0.003

(0.439-0.527)

(0.494-0.521)

(0.455-0.545)

(0.003-0.005)

(0.001-0.002)

(0.002-0.005)

0.585

0.599

0.611

0.004

0.002

0.005

(0.536-0.636)

(0.570-0.634)

(0.561-0.662)

(0.003-0.004)

(0.002-0.003)

(0.003-0.007)

0.747

0.725

0.761

0.004

0.002

0.004

(0.710-0.785)

(0.693-0.757)

(0.726-0.799)

(0.003-0.005)

(0.001-0.002)

(0.002-0.006)

0.586

0.559

0.583

0.036

0.005

0.011

(0.574-0.596)

(0.551-0.567)

(0.572-0.593)

(0.035-0.038)

(0.003-0.006)

(0.008-0.015)

0.627

0.592

0.631

0.034

0.003

0.007

(0.616-0.640)

(0.581-0.601)

(0.619-0.643)

(0.033-0.035)

(0.002-0.004)

(0.005-0.011)

0.689

0.642

0.717

0.036

0.004

0.008

(0.676-0.702)

(0.631-0.652)

(0.705-0.729)

(0.034-0.037)

(0.002-0.005)

(0.005-0.012)

0.661

0.613

0.644

0.026

0.006

0.014

(0.647-0.677)

(0.601-0.625)

(0.632-0.658)

(0.024-0.027)

(0.004-0.008)

(0.008-0.022)

0.663

0.625

0.683

0.024

0.005

0.014

(0.645-0.680)

(0.611-0.638)

(0.664-0.700)

(0.022-0.025)

(0.004-0.006)

(0.007-0.022)

0.836

0.767

0.853

0.024

0.004

0.014

(0.822-0.847)

(0.755-0.778)

(0.839-0.862)

(0.023-0.026)

(0.002-0.005)

(0.008-0.020)

0.689

0.645

0.669

0.012

0.002

0.005

(0.669-0.713)

(0.624-0.665)

(0.648-0.692)

(0.011-0.013)

(0.001-0.003)

(0.003-0.008)

0.692

0.644

0.694

0.012

0.002

0.007

(0.668-0.717)

(0.629-0.666)

(0.672-0.720)

(0.011-0.013)

(0.001-0.003)

(0.004-0.011)

0.762

0.701

0.789

0.012

0.004

0.014

(0.736-0.789)

(0.683-0.721)

(0.764-0.814)

(0.011-0.013)

(0.003-0.005)

(0.009-0.019)

0.672

0.633

0.671

0.013

0.001

0.003

(0.658-0.689)

(0.617-0.647)

(0.656-0.687)

(0.012-0.013)

(0.001-0.002)

(0.002-0.005)

0.707

0.660

0.705

0.012

0.002

0.007

(0.694-0.724)

(0.648-0.675)

(0.691-0.721)

(0.012-0.013)

(0.001-0.003)

(0.004-0.010)

0.788

0.718

0.789

0.012

0.002

0.005

(0.777-0.800)

(0.705-0.731)

(0.778-0.800)

(0.012-0.013)

(0.001-0.002)

(0.003-0.009)

0.571

0.557

0.559

0.002

0.001

0.002

(0.518-0.628)

(0.509-0.608)

(0.507-0.614)

(0.002-0.002)

(0.001-0.002)

(0.002-0.003)

Continued on next page

Last Progress A POLLO Sleep Apnea

Age-Sex Last Progress A POLLO

0.590

0.564

0.623

0.002

0.008

(0.527-0.663)

(0.519-0.609)

(0.561-0.695)

(0.001-0.002)

(0.006-0.010)

0.727

0.671

0.734

0.002

0.010

(0.658-0.779)

(0.622-0.709)

(0.667-0.785)

(0.002-0.002)

(0.001-0.002)

(0.007-0.012)

0.605

0.599

0.598

0.046

0.012

0.022

(0.595-0.613)

(0.592-0.606)

(0.589-0.606)

(0.045-0.048)

(0.011-0.014)

(0.016-0.027)

0.643

0.606

0.643

0.045

0.003

0.009

(0.629-0.652)

(0.598-0.614)

(0.631-0.651)

(0.043-0.046)

(0.002-0.005)

(0.004-0.014)

0.758

0.694

0.768

0.045

0.004

0.015

(0.748-0.765)

(0.687-0.702)

(0.758-0.774)

(0.043-0.046)

(0.002-0.005)

(0.007-0.021)

Extended Data Table 8: Definition of disease progression prediction tasks. We evaluated TTE performance on 78 disease progression tasks. τ is the threshold duration. Incidence refers to the percentage of patients in the test set who reach the endpoint by the specified τ . The Snapshot Event and Endpoint Event columns specify all ICD-10 codes used to identify the disease indicated in the Task column. For some tasks, where the Endpoint Event is the administration of a medication, we provide the list of medication names used to identify the endpoint. Category

Task

Incid.

Snapshot Event

Endpoint Event

Cardiovascular & Circulatory

Aortic Stenosis Mortality

7,352

14.87%

death

Cardiovascular & Circulatory

Atrial Fibrillation to Ischemic Stroke Chronic Ischemic Heart Disease to Acute Myocardial Infarction Chronic Ischemic Heart Disease to Heart Failure Heart Failure Mortality Heart Failure to Acute Kidney Failure Hypertension to Atrial Fibrillation Hypertension to Chronic Kidney Disease Stage 3+ Hypertension to Heart Failure Paroxysmal to Persistent Afib Stroke Mortality Benign Prostatic Hyperplasia to Urinary Retention Diabetes Mellitus to Diabetic Retinopathy

23,886

2.87%

I06.0, I06.2, I35.0, I35.2, Q23.0, Q25.3 I48

27,979

2.66%

I25

I21

31,812

7.30%

I25

I50

3y 1y

20,436 15,872

25.79% 3.58%

I50 I50

death N17

139,256

5.72%

I10

I48

141,287

3.96%

I10

137,799

3.97%

I10

N18.3, N18.6 I50

8,966

2.02%

I48.0

I48.1

3y 3y

15,508 17,782

14.26% 2.49%

I60, I61, I62, I63 N40

death R33

48,673

3.12%

E08, E09, E10, E11, E13

E08.31, E08.32, E08.33, E08.34, E08.35, E09.31, E09.32, E09.33, E09.34, E09.35, E10.31, E10.32, E10.33, E10.34, E10.35, E11.31, E11.32, E11.33, E11.34, E11.35, E13.31, E13.32, E13.33, E13.34, E13.35 I70.2, I70.3, I70.4, I70.5, I70.6, I70.7, I73.9 E11

Cardiovascular & Circulatory

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive

I63

N18.4,

N18.5,

Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive

Diabetes Mellitus to Peripheral Artery Disease Gestational Diabetes to Type 2 Diabetes Non-Proliferative to Proliferative Diabetic Retinopathy

47,969

2.46%

E08, E09, E10, E11, E13

2,705

7.80%

O24.4

5,922

2.77%

E08.35, E09.35, E10.35, E11.35, E13.35

Obesity to Type 2 Diabetes

59,452

8.95%

Prediabetes to Type 2 Diabetes Type 2 Diabetes to Diabetic Foot Ulcer Type 2 Diabetes to Diabetic Nephropathy Type 2 Diabetes to Diabetic Neuropathy Type 2 Diabetes to Insulin Use

18,161

7.27%

E08.32, E08.33, E08.34, E09.32, E09.33, E09.34, E10.32, E10.33, E10.34, E11.32, E11.33, E11.34, E13.32, E13.33, E13.34 E66.0, E66.1, E66.2, E66.8, E66.9 R73.03

43,890

0.54%

E11

44,198

1.28%

E11

E10.621, E11.621, E13.621 E10.21, E11.21, E13.21

45,905

4.02%

E11

E10.4, E11.4, E13.4

58,642

7.91%

E11

Gastrointestinal, Hepatic & Renal

CKD Stage 2 to Stage 3

2,218

20.82%

N18.2

insulin, Humalog, Admelog, Lyumjev, Novolog, Fiasp, Merilog, Trurapi, NovoRapid, Apidra, Humulin, Novolin, Myxredlin, Lantus, Basaglar, Toujeo, Semglee, Rezvoglar, Levemir, Tresiba N18.3

E11 E11

Continued on next page

Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal

CKD Stage 3 to Stage 4+ Chronic Hepatitis or Fatty Liver to Cirrhosis

3y 5y

12,229 12,482

4.05% 3.92%

Gastrointestinal, Hepatic & Renal

Cirrhosis to Esophageal Varices Hepatitis C to Cirrhosis NAFLD to Liver Fibrosis NAFLD to NASH NASH to Cirrhosis Glaucoma to Blindness Osteoporosis to Bone Fracture Psoriasis to Psoriatic Arthritis Rheumatoid Arthritis to Interstitial Lung Disease Rheumatoid Arthritis to Second-Line Therapy

2,928

5y 5y 3y 3y 5y 3y

Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Hematology & Sensory Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic

Neurology & Psychiatry Neurology & Psychiatry

Neurology & Psychiatry Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers)

Episodic to Chronic Migraine Mild Cognitive Impairment to Alzheimer’s Disease Mild Cognitive Impairment to Dementia Acute Lymphocytic Leukemia Mortality Acute Myeloid Leukemia Mortality All Leukemia Mortality Barrett’s Esophagus to Esophageal Cancer Bladder Cancer Metastasis Bladder Cancer Mortality Breast Cancer Metastasis Breast Cancer Mortality Breast Cancer Palliative Care Chronic Lymphocytic Leukemia Mortality Chronic Myeloid Leukemia Mortality Cirrhosis to Liver Cancer Colorectal Cancer Metastasis Colorectal Cancer Mortality GERD to Barrett’s Esophagus Inflammatory Bowel Disease to Colorectal Cancer Kidney Cancer Metastasis Kidney Cancer Mortality Liver Cancer Metastasis Liver Cancer Mortality Lung Cancer Metastasis Lung Cancer Mortality Melanoma Metastasis

N18.4, N18.5, N18.6 K70.3, K71.7, K74

8.32%

N18.3 B18, K70.1, K71.3, K71.4, K71.5, K71.6, K73, K75, K70.0, K70.2, K76.0 K70.3, K71.7, K74

3,716 16,897 17,794 1,955 19,281 20,813

4.67% 0.48% 0.45% 4.36% 0.94% 0.55%

B17.1, B18.2, B19.2 K76.0 K76.0 K75.81 H40 M81

K70.3, K71.7, K74 K74.0 K75.81 K70.3, K71.7, K74 H54 M80, M84

7,791

3.87%

L40

L40.5

5,667

1.83%

M05, M06

J84

6,232

3.68%

M05, M06

27,580

0.11%

G43

Adalimumab, Humira, Amjevita, Cyltezo, Hyrimoz, Idacio, Abrilada, Hadlima, Etanercept, Enbrel, Erelzi, Eticovo, Infliximab, Remicade, Inflectra, Renflexis, Ixifi, Avsola, Golimumab, Simponi, Certolizumab pegol, Cimzia, Tocilizumab, Actemra, Sarilumab, Kevzara, Rituximab, Rituxan, Truxima, Ruxience, Riabni, Abatacept, Orencia, Tofacitinib, Xeljanz, Baricitinib, Olumiant, Upadacitinib, Rinvoq G43.7

1,650

10.53%

F06.7, G31.84

G30

1,803

16.72%

F06.7, G31.84

2,053

17.03%

C91.0

F01, F03, G30, G31.0, G31.83 death

1,029

48.19%

C92.0

death

3y 5y

4,357 7,271

24.73% 0.79%

C91, C92, C93, C94, C95 K22.7

death C15

3y 3y 5y 3y 5y

7,049 3,562 14,383 17,752 14,460

4.09% 20.60% 4.98% 7.18% 2.64%

C67 C67 C50 C50 C50

C77, C78, C79 death C77, C78, C79 death Z51.5

1,855

19.25%

C91.1

death

2,287

19.71%

C92.1, C92.2

death

3y 3y

2,855 4,410

3.79% 7.85%

K70.3, K71.7, K74 C18, C19, C20

C22 C77, C78, C79

6,163

22.09%

C18, C19, C20

death

61,891

1.42%

K21

K22.7

12,088

0.44%

K50, K51, K52.3

C18, C19, C20

2y 3y 2y 1y 2y 2y 2y

2,031 2,609 2,673 1,416 6,208 8,110 4,315

6.97% 18.17% 5.51% 33.23% 10.95% 35.88% 3.68%

C64, C65 C64, C65 C22 C22 C34 C34 C43

C77, C78, C79 death C77, C78, C79 death C77, C78, C79 death C77, C78, C79

I85

Continued on next page

Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious

Melanoma Mortality Myelodysplastic Syndrome to Leukemia Ovarian Cancer Metastasis Ovarian Cancer Mortality Pancreatic Cancer Metastasis Pancreatic Cancer Mortality Prostate Cancer Metastasis Prostate Cancer Mortality Thyroid Cancer Metastasis Thyroid Cancer Mortality Acute COVID-19 to Long COVID Asthma to Exacerbation Requiring Oral Corticosteroids

COPD to Chronic Respiratory Failure COPD to Exacerbation COPD to Pulmonary Hypertension COVID-19 Mortality Pulmonary Fibrosis to Pulmonary Hypertension

3y 2y

5,262 3,608

8.66% 11.08%

C43 D46

death C91, C92, C93, C94, C95

3y 3y 2y

3,823 2,010 4,454

5.44% 33.91% 12.06%

C56 C56 C25

C77, C78, C79 death C77, C78, C79

2,299

64.86%

C25

death

5y 5y 5y 5y 1y

9,662 12,640 6,341 2,942 16,604

4.32% 15.60% 2.20% 6.11% 0.55%

C61 C61 C73 C73 U07.1

C77, C78, C79 death C77, C78, C79 death U09.9

57,743

2.82%

J45

12,590

1.68%

J44

Prednisone, Deltasone, Rayos, Sterapred, Prednisolone, Orapred, Prelone, Millipred, Pediapred, Methylprednisolone, Medrol, Hydrocortisone, Cortef, Hydrocort, Dexamethasone, Decadron, DexPak, Betamethasone, Celestone J96.1

3y 3y

12,424 12,446

2.97% 2.15%

J44 J44

J44.0, J44.1 I27.2

60d 3y

14,011 3,190

0.18% 8.32%

U07.1 J84.1

death I27.2

Extended Data Table 9: Evaluation results of disease progression prediction tasks. We report timedependent performance metrics computed at τ on the test set for 78 disease progression tasks. τ is the threshold duration. Arrows indicate whether higher (↑) or lower (↓) values are better. ICI, Integrated Calibration Index; MCE, Maximum Calibration Error. 10−4 is the default value of the penalizer (λ) for Cox proportional hazards; if convergence cannot be achieved then λ is increased to 1e-3 (∗ ) or 1e-2 (∗∗ ). Mean and 95% CI reported over 100 bootstraps of the test set. Task

Model

Aortic Stenosis Mortality

Age-Sex Last Progress A POLLO

Atrial Fibrillation Stroke

Ischemic

Age-Sex Last Progress A POLLO

Chronic Ischemic Heart Disease to Acute Myocardial Infarction

Age-Sex Last Progress A POLLO

Chronic Ischemic Heart Disease to Heart Failure

Age-Sex Last Progress A POLLO

Heart Failure Mortality

Age-Sex Last Progress A POLLO

Heart Failure to Acute Kidney Failure

Age-Sex Last Progress A POLLO

Hypertension to Atrial Fibrillation

Age-Sex Last Progress A POLLO

Hypertension to Chronic Kidney Disease Stage 3+

Age-Sex Last Progress A POLLO

Hypertension to Heart Failure

Age-Sex Last Progress A POLLO

Paroxysmal to Persistent Afib

Age-Sex

AUC (↑)

Bal. Acc. (↑)

C-Index (↑)

Brier Score (↓)

ICI (↓)

MCE (↓)

0.722

0.653

0.699

0.099

0.041

0.092

(0.684-0.755)

(0.618-0.687)

(0.665-0.731)

(0.091-0.108)

(0.029-0.053)

(0.061-0.135)

0.648

0.618

0.643

0.135

0.052

0.172

(0.598-0.702)

(0.575-0.670)

(0.597-0.691)

(0.119-0.147)

(0.035-0.070)

(0.103-0.262)

0.825

0.738

0.801

0.088

0.038

0.103

(0.797-0.855)

(0.701-0.772)

(0.775-0.827)

(0.078-0.096)

(0.027-0.051)

(0.058-0.169)

0.615

0.561

0.594

0.022

0.006

0.017

(0.572-0.653)

(0.527-0.591)

(0.552-0.632)

(0.019-0.025)

(0.004-0.008)

(0.011-0.024)

0.562

0.549

0.561

0.024

0.015

0.048

(0.500-0.618)

(0.497-0.595)

(0.499-0.616)

(0.020-0.028)

(0.010-0.019)

(0.028-0.068)

0.677

0.630

0.658

0.022

0.008

0.034

(0.644-0.712)

(0.598-0.668)

(0.628-0.693)

(0.019-0.025)

(0.006-0.011)

(0.019-0.049)

0.485

0.497

0.496

0.021

0.005

0.012

(0.449-0.527)

(0.493-0.503)

(0.461-0.537)

(0.018-0.024)

(0.004-0.007)

(0.007-0.018)

0.512

0.511

0.522

0.026

0.017

0.062

(0.459-0.578)

(0.455-0.560)

(0.472-0.588)

(0.022-0.031)

(0.013-0.021)

(0.042-0.080)

0.604

0.582

0.601

0.021

0.006

0.016

(0.567-0.644)

(0.557-0.618)

(0.565-0.640)

(0.018-0.024)

(0.004-0.008)

(0.008-0.024)

0.649

0.615

0.636

0.055

0.011

0.034

(0.628-0.669)

(0.594-0.635)

(0.616-0.656)

(0.052-0.059)

(0.007-0.016)

(0.018-0.055)

0.630

0.606

0.621

0.070

0.021

0.065

(0.600-0.662)

(0.581-0.635)

(0.591-0.653)

(0.064-0.076)

(0.014-0.027)

(0.039-0.091)

0.768

0.699

0.748

0.052

0.009

0.022

(0.743-0.786)

(0.679-0.715)

(0.723-0.766)

(0.049-0.056)

(0.005-0.012)

(0.012-0.041)

0.677

0.627

0.647

0.157

0.019

0.047

(0.660-0.698)

(0.612-0.642)

(0.632-0.664)

(0.150-0.162)

(0.012-0.027)

(0.030-0.069)

0.622

0.603

0.622

0.186

0.042

0.127

(0.593-0.646)

(0.582-0.627)

(0.601-0.640)

(0.178-0.194)

(0.030-0.054)

(0.081-0.183)

0.766

0.694

0.732

0.142

0.016

0.043

(0.753-0.781)

(0.682-0.711)

(0.722-0.745)

(0.135-0.146)

(0.011-0.023)

(0.023-0.077)

0.525

0.517

0.518

0.031

0.008

0.019

(0.480-0.565)

(0.484-0.555)

(0.473-0.557)

(0.026-0.035)

(0.005-0.012)

(0.011-0.028)

0.562

0.551

0.557

0.046

0.020

0.055

(0.518-0.604)

(0.512-0.588)

(0.513-0.597)

(0.041-0.052)

(0.014-0.026)

(0.036-0.074)

0.724

0.650

0.714

0.030

0.008

0.021

(0.698-0.758)

(0.619-0.682)

(0.687-0.747)

(0.026-0.035)

(0.005-0.012)

(0.011-0.033)

0.758

0.696

0.738

0.042

0.006

0.016

(0.748-0.768)

(0.688-0.705)

(0.728-0.749)

(0.041-0.044)

(0.004-0.007)

(0.010-0.024)

0.644

0.591

0.651

0.052

0.009

0.026

(0.624-0.665)

(0.569-0.608)

(0.630-0.669)

(0.049-0.056)

(0.005-0.013)

(0.016-0.039)

0.796

0.721

0.773

0.041

0.005

0.014

(0.787-0.804)

(0.710-0.732)

(0.765-0.782)

(0.039-0.042)

(0.003-0.007)

(0.007-0.022)

0.686

0.639

0.668

0.030

0.005

0.028

(0.673-0.697)

(0.630-0.651)

(0.656-0.679)

(0.029-0.032)

(0.004-0.007)

(0.021-0.036)

0.629

0.597

0.631

0.050

0.008

0.024

(0.602-0.654)

(0.575-0.619)

(0.608-0.654)

(0.047-0.053)

(0.005-0.012)

(0.013-0.040)

0.820

0.744

0.798

0.029

0.006

0.018

(0.810-0.830)

(0.734-0.756)

(0.788-0.808)

(0.027-0.030)

(0.004-0.007)

(0.008-0.029)

0.755

0.690

0.737

0.030

0.002

0.006

(0.741-0.769)

(0.675-0.703)

(0.724-0.752)

(0.028-0.031)

(0.001-0.003)

(0.003-0.014)

0.700

0.652

0.699

0.037

0.010

0.051

(0.676-0.727)

(0.628-0.675)

(0.675-0.726)

(0.034-0.040)

(0.007-0.012)

(0.030-0.069)

0.856

0.780

0.837

0.028

0.004

0.012

(0.845-0.867)

(0.770-0.790)

(0.826-0.848)

(0.027-0.029)

(0.002-0.005)

(0.005-0.023)

0.517

0.519

0.494

0.015

0.010

0.022

(0.449-0.570)

(0.472-0.562)

(0.430-0.552)

(0.011-0.019)

(0.007-0.013)

(0.015-0.029)

Continued on next page

Last Progress A POLLO Stroke Mortality

Age-Sex Last Progress A POLLO

Benign Prostatic Hyperplasia to Urinary Retention

Age-Sex Last Progress A POLLO

Diabetes Mellitus Retinopathy

Diabetic

Age-Sex Last Progress A POLLO

Diabetes Mellitus to Peripheral Artery Disease

Age-Sex Last Progress A POLLO

Gestational Diabetes to Type 2 Diabetes

Age-Sex Last Progress∗ A POLLO

Non-Proliferative to Proliferative Diabetic Retinopathy

Age-Sex Last Progress A POLLO

Obesity to Type 2 Diabetes

Age-Sex Last Progress A POLLO

Prediabetes to Type 2 Diabetes

Age-Sex Last Progress A POLLO

Type 2 Diabetes to Diabetic Foot Ulcer

Age-Sex Last Progress A POLLO

Type 2 Diabetes Nephropathy

Diabetic

Age-Sex Last Progress A POLLO

Type 2 Diabetes to Diabetic Neuropathy

Age-Sex

0.521

0.518

0.529

0.022

0.023

0.106

(0.451-0.598)

(0.445-0.601)

(0.459-0.607)

(0.015-0.027)

(0.017-0.029)

(0.088-0.126)

0.603

0.553

0.596

0.015

0.014

0.068

(0.514-0.677)

(0.502-0.612)

(0.510-0.668)

(0.011-0.020)

(0.011-0.018)

(0.047-0.083)

0.723

0.660

0.697

0.094

0.020

0.054

(0.699-0.745)

(0.636-0.680)

(0.673-0.717)

(0.088-0.101)

(0.012-0.031)

(0.027-0.087)

0.722

0.668

0.704

0.124

0.030

0.074

(0.688-0.750)

(0.636-0.694)

(0.679-0.729)

(0.114-0.133)

(0.020-0.040)

(0.048-0.117)

0.837

0.756

0.808

0.081

0.014

0.040

(0.821-0.855)

(0.738-0.776)

(0.793-0.825)

(0.074-0.087)

(0.008-0.023)

(0.019-0.079)

0.617

0.586

0.602

0.021

0.006

0.013

(0.576-0.659)

(0.539-0.631)

(0.563-0.645)

(0.017-0.024)

(0.003-0.008)

(0.008-0.021)

0.490

0.491

0.032

0.022

0.079

(0.414-0.559)

(0.428-0.560)

(0.415-0.559)

(0.025-0.040)

(0.017-0.029)

(0.060-0.097)

0.688

0.637

0.668

0.021

0.007

0.023

(0.652-0.728)

(0.600-0.665)

(0.630-0.707)

(0.018-0.024)

(0.004-0.010)

(0.010-0.038)

0.537

0.525

0.524

0.024

0.004

0.009

(0.510-0.566)

(0.508-0.539)

(0.497-0.552)

(0.022-0.026)

(0.002-0.006)

(0.006-0.013)

0.524

0.531

0.526

0.022

0.012

0.037

(0.453-0.585)

(0.488-0.578)

(0.456-0.586)

(0.018-0.026)

(0.008-0.015)

(0.021-0.054)

0.650

0.598

0.657

0.023

0.004

0.012

(0.626-0.670)

(0.574-0.624)

(0.635-0.678)

(0.021-0.026)

(0.003-0.006)

(0.007-0.018)

0.693

0.634

0.671

0.019

0.005

0.014

(0.669-0.721)

(0.608-0.662)

(0.648-0.698)

(0.016-0.020)

(0.003-0.007)

(0.008-0.023)

0.604

0.569

0.622

0.023

0.014

0.059

(0.547-0.664)

(0.527-0.613)

(0.571-0.674)

(0.018-0.027)

(0.011-0.018)

(0.037-0.076)

0.765

0.691

0.748

0.018

0.004

0.010

(0.742-0.795)

(0.662-0.719)

(0.726-0.778)

(0.016-0.020)

(0.002-0.005)

(0.005-0.018)

0.446

0.452

0.470

0.052

0.032

0.078

(0.333-0.547)

(0.360-0.536)

(0.351-0.569)

(0.035-0.067)

(0.020-0.048)

(0.051-0.135)

0.561

0.551

0.537

0.074

0.090

0.243

(0.420-0.660)

(0.445-0.641)

(0.397-0.620)

(0.047-0.104)

(0.061-0.118)

(0.164-0.349)

0.596

0.604

0.611

0.053

0.042

0.147

(0.492-0.708)

(0.520-0.680)

(0.517-0.717)

(0.040-0.068)

(0.025-0.060)

(0.061-0.220)

0.541

0.530

0.563

0.021

0.009

0.024

(0.467-0.627)

(0.468-0.587)

(0.491-0.645)

(0.015-0.026)

(0.005-0.014)

(0.013-0.048)

0.589

0.604

0.585

0.022

0.016

0.045

(0.466-0.688)

(0.523-0.684)

(0.464-0.686)

(0.013-0.029)

(0.009-0.022)

(0.024-0.076)

0.629

0.557

0.630

0.020

0.011

0.028

(0.542-0.721)

(0.497-0.619)

(0.546-0.719)

(0.015-0.027)

(0.007-0.017)

(0.018-0.050)

0.698

0.652

0.679

0.061

0.016

0.035

(0.686-0.709)

(0.641-0.663)

(0.668-0.690)

(0.058-0.064)

(0.013-0.019)

(0.025-0.045)

0.590

0.566

0.599

0.060

0.014

0.037

(0.561-0.618)

(0.539-0.589)

(0.572-0.627)

(0.056-0.066)

(0.010-0.018)

(0.021-0.056)

0.744

0.681

0.739

0.060

0.009

0.022

(0.729-0.758)

(0.666-0.692)

(0.725-0.752)

(0.057-0.062)

(0.006-0.012)

(0.012-0.035)

0.561

0.544

0.562

0.049

0.012

0.027

(0.532-0.595)

(0.516-0.570)

(0.532-0.595)

(0.045-0.053)

(0.008-0.016)

(0.017-0.042)

0.526

0.528

0.532

0.052

0.019

0.063

(0.482-0.572)

(0.494-0.569)

(0.491-0.577)

(0.046-0.058)

(0.014-0.026)

(0.044-0.085)

0.679

0.629

0.671

0.048

0.013

0.045

(0.652-0.717)

(0.600-0.659)

(0.644-0.709)

(0.043-0.054)

(0.009-0.018)

(0.025-0.075)

0.665

0.656

0.670

0.004

0.002

0.005

(0.602-0.722)

(0.603-0.697)

(0.607-0.728)

(0.003-0.005)

(0.001-0.003)

(0.003-0.010)

0.610

0.612

0.613

0.006

0.034

(0.490-0.770)

(0.493-0.722)

(0.493-0.768)

(0.004-0.008)

(0.020-0.044)

0.814

0.704

0.805

0.004

0.002

0.008

(0.757-0.865)

(0.639-0.761)

(0.746-0.856)

(0.003-0.005)

(0.002-0.004)

(0.004-0.014)

0.550

0.526

0.539

0.010

0.003

0.008

(0.509-0.592)

(0.483-0.574)

(0.499-0.580)

(0.008-0.011)

(0.002-0.005)

(0.004-0.013)

0.534

0.541

0.515

0.010

0.009

0.026

(0.442-0.626)

(0.460-0.629)

(0.426-0.602)

(0.007-0.014)

(0.006-0.012)

(0.015-0.038)

0.663

0.606

0.658

0.010

0.003

0.008

(0.624-0.705)

(0.571-0.640)

(0.619-0.701)

(0.009-0.011)

(0.002-0.005)

(0.005-0.012)

0.588

0.576

0.568

0.030

0.006

0.014

(0.565-0.614)

(0.555-0.600)

(0.546-0.593)

(0.027-0.032)

(0.004-0.008)

(0.008-0.021)

Continued on next page

Last Progress A POLLO Type 2 Diabetes to Insulin Use

Age-Sex Last Progress A POLLO

CKD Stage 2 to Stage 3

Age-Sex Last Progress A POLLO

CKD Stage 3 to Stage 4+

Age-Sex Last Progress A POLLO

Chronic Hepatitis or Fatty Liver to Cirrhosis

Age-Sex Last Progress A POLLO

Cirrhosis to Esophageal Varices

Age-Sex Last Progress A POLLO

Hepatitis C to Cirrhosis

Age-Sex Last Progress∗ A POLLO

NAFLD to Liver Fibrosis

Age-Sex Last Progress A POLLO

NAFLD to NASH

Age-Sex Last Progress A POLLO

NASH to Cirrhosis

Age-Sex Last Progress A POLLO

Glaucoma to Blindness

Age-Sex Last Progress∗ A POLLO

Osteoporosis to Bone Fracture

Age-Sex Last Progress A POLLO

0.582

0.562

0.587

0.033

0.015

0.040

(0.528-0.633)

(0.519-0.610)

(0.534-0.633)

(0.027-0.039)

(0.010-0.019)

(0.021-0.063)

0.682

0.626

0.664

0.029

0.005

0.014

(0.657-0.708)

(0.599-0.649)

(0.639-0.690)

(0.027-0.032)

(0.003-0.008)

(0.009-0.022)

0.556

0.547

0.545

0.064

0.011

0.025

(0.539-0.576)

(0.534-0.564)

(0.529-0.564)

(0.061-0.067)

(0.008-0.015)

(0.016-0.035)

0.622

0.601

0.624

0.122

0.021

0.052

(0.594-0.645)

(0.579-0.622)

(0.600-0.644)

(0.115-0.131)

(0.013-0.030)

(0.025-0.090)

0.796

0.727

0.769

0.058

0.009

0.019

(0.780-0.811)

(0.713-0.743)

(0.755-0.783)

(0.056-0.061)

(0.006-0.013)

(0.012-0.031)

0.622

0.592

0.604

0.128

0.049

0.120

(0.565-0.679)

(0.540-0.640)

(0.546-0.657)

(0.107-0.150)

(0.028-0.069)

(0.068-0.200)

0.413

0.459

0.433

0.213

0.184

0.419

(0.326-0.488)

(0.397-0.523)

(0.371-0.495)

(0.179-0.247)

(0.136-0.232)

(0.335-0.529)

0.705

0.649

0.637

0.123

0.075

0.227

(0.658-0.751)

(0.598-0.698)

(0.596-0.682)

(0.102-0.145)

(0.048-0.098)

(0.116-0.358)

0.587

0.575

0.568

0.030

0.011

0.028

(0.543-0.637)

(0.540-0.611)

(0.524-0.615)

(0.026-0.035)

(0.007-0.015)

(0.018-0.037)

0.587

0.573

0.572

0.037

0.021

0.075

(0.535-0.642)

(0.526-0.623)

(0.523-0.627)

(0.030-0.044)

(0.015-0.027)

(0.045-0.100)

0.777

0.703

0.760

0.029

0.008

0.024

(0.736-0.815)

(0.670-0.730)

(0.721-0.797)

(0.025-0.033)

(0.005-0.011)

(0.011-0.042)

0.646

0.603

0.641

0.027

0.008

0.021

(0.595-0.686)

(0.550-0.640)

(0.592-0.680)

(0.023-0.031)

(0.005-0.011)

(0.010-0.038)

0.548

0.557

0.551

0.035

0.029

0.093

(0.479-0.647)

(0.490-0.629)

(0.478-0.649)

(0.027-0.043)

(0.023-0.037)

(0.056-0.125)

0.753

0.691

0.747

0.026

0.014

0.051

(0.699-0.792)

(0.640-0.735)

(0.696-0.787)

(0.022-0.031)

(0.009-0.018)

(0.026-0.079)

0.549

0.541

0.568

0.059

0.028

0.065

(0.483-0.622)

(0.481-0.604)

(0.503-0.637)

(0.047-0.075)

(0.016-0.039)

(0.033-0.109)

0.523

0.519

0.533

0.081

0.074

0.241

(0.397-0.642)

(0.421-0.619)

(0.423-0.642)

(0.062-0.104)

(0.050-0.096)

(0.115-0.342)

0.706

0.669

0.705

0.057

0.028

0.081

(0.623-0.773)

(0.606-0.726)

(0.627-0.769)

(0.046-0.072)

(0.014-0.040)

(0.040-0.133)

0.733

0.689

0.726

0.035

0.020

0.054

(0.676-0.782)

(0.622-0.744)

(0.671-0.769)

(0.027-0.047)

(0.013-0.029)

(0.027-0.094)

0.516

0.445

0.518

0.050

0.148

(0.296-0.708)

(0.328-0.586)

(0.304-0.699)

(0.028-0.074)

(0.029-0.076)

(0.062-0.259)

0.728

0.681

0.726

0.035

0.021

0.075

(0.657-0.794)

(0.622-0.744)

(0.659-0.789)

(0.027-0.044)

(0.013-0.029)

(0.034-0.130)

0.489

0.507

0.487

0.003

0.002

0.006

(0.384-0.593)

(0.416-0.594)

(0.383-0.590)

(0.002-0.004)

(0.001-0.003)

(0.003-0.010)

0.529

0.540

0.545

0.008

0.006

0.018

(0.399-0.644)

(0.475-0.615)

(0.414-0.660)

(0.005-0.011)

(0.004-0.009)

(0.010-0.031)

0.719

0.704

0.693

0.003

0.009

(0.642-0.797)

(0.627-0.779)

(0.617-0.769)

(0.002-0.004)

(0.005-0.015)

0.516

0.547

0.519

0.004

0.003

0.005

(0.439-0.591)

(0.488-0.607)

(0.440-0.593)

(0.002-0.005)

(0.002-0.004)

(0.004-0.009)

0.570

0.496

0.570

0.004

0.005

0.021

(0.467-0.676)

(0.420-0.581)

(0.464-0.676)

(0.003-0.006)

(0.004-0.006)

(0.018-0.024)

0.590

0.549

0.591

0.004

0.003

0.009

(0.493-0.689)

(0.487-0.626)

(0.494-0.689)

(0.003-0.005)

(0.002-0.004)

(0.005-0.013)

0.670

0.599

0.669

0.032

0.021

0.058

(0.578-0.774)

(0.496-0.697)

(0.581-0.769)

(0.022-0.042)

(0.011-0.032)

(0.025-0.097)

0.647

0.455

0.646

0.046

0.048

0.142

(0.520-0.747)

(0.434-0.470)

(0.529-0.738)

(0.030-0.063)

(0.031-0.067)

(0.091-0.231)

0.763

0.717

0.765

0.030

0.019

0.071

(0.681-0.835)

(0.630-0.797)

(0.687-0.836)

(0.018-0.041)

(0.011-0.028)

(0.034-0.127)

0.531

0.500

0.519

0.007

0.004

0.011

(0.419-0.622)

(0.421-0.569)

(0.406-0.611)

(0.005-0.009)

(0.003-0.006)

(0.005-0.019)

0.558

0.578

0.556

0.008

0.023

(0.408-0.709)

(0.459-0.670)

(0.401-0.708)

(0.005-0.011)

(0.010-0.037)

0.676

0.625

0.663

0.008

0.034

(0.618-0.746)

(0.558-0.695)

(0.607-0.733)

(0.006-0.010)

(0.021-0.045)

0.675

0.585

0.661

0.005

0.003

0.006

(0.590-0.776)

(0.511-0.648)

(0.573-0.764)

(0.003-0.006)

(0.002-0.004)

(0.004-0.012)

0.503

0.496

0.505

0.013

0.034

(0.395-0.613)

(0.442-0.559)

(0.399-0.615)

(0.009-0.016)

(0.010-0.017)

(0.020-0.049)

0.770

0.700

0.753

0.005

0.002

0.010

(0.662-0.846)

(0.604-0.758)

(0.645-0.830)

(0.004-0.006)

(0.001-0.004)

(0.004-0.019)

Continued on next page

Psoriasis to Psoriatic Arthritis

Age-Sex Last Progress A POLLO

Rheumatoid Arthritis to Interstitial Lung Disease

Age-Sex Last Progress∗∗ A POLLO

Rheumatoid Arthritis to SecondLine Therapy

Age-Sex Last Progress A POLLO

Episodic to Chronic Migraine

Age-Sex Last Progress A POLLO

Mild Cognitive Impairment to Alzheimer’s Disease

Age-Sex Last Progress A POLLO

Mild Cognitive Impairment to Dementia

Age-Sex Last Progress A POLLO

Acute Lymphocytic Mortality

Leukemia

Age-Sex Last Progress A POLLO

Acute Myeloid Leukemia Mortality

Age-Sex Last Progress A POLLO

All Leukemia Mortality

Age-Sex Last Progress A POLLO

Barrett’s Esophagus to Esophageal Cancer

Age-Sex Last Progress A POLLO

Bladder Cancer Metastasis

Age-Sex Last Progress A POLLO

Bladder Cancer Mortality

Age-Sex

0.512

0.502

0.517

0.029

0.011

0.027

(0.461-0.562)

(0.462-0.541)

(0.468-0.566)

(0.023-0.035)

(0.007-0.016)

(0.016-0.042)

0.580

0.574

0.584

0.047

0.045

0.136

(0.505-0.647)

(0.512-0.652)

(0.510-0.657)

(0.028-0.065)

(0.029-0.059)

(0.090-0.198)

0.544

0.520

0.563

0.030

0.022

0.067

(0.490-0.601)

(0.472-0.571)

(0.513-0.619)

(0.025-0.036)

(0.016-0.027)

(0.043-0.091)

0.691

0.664

0.662

0.014

0.011

0.038

(0.613-0.771)

(0.608-0.723)

(0.585-0.737)

(0.009-0.018)

(0.007-0.015)

(0.017-0.059)

0.424

0.451

0.466

0.027

0.029

0.101

(0.308-0.572)

(0.408-0.517)

(0.345-0.611)

(0.016-0.039)

(0.018-0.043)

(0.060-0.191)

0.613

0.535

0.619

0.014

0.012

0.035

(0.476-0.713)

(0.438-0.618)

(0.485-0.718)

(0.010-0.018)

(0.007-0.017)

(0.019-0.060)

0.490

0.484

0.498

0.031

0.012

0.032

(0.440-0.546)

(0.433-0.540)

(0.449-0.555)

(0.024-0.037)

(0.007-0.019)

(0.016-0.059)

0.520

0.507

0.516

0.076

0.065

0.200

(0.433-0.610)

(0.449-0.557)

(0.433-0.604)

(0.056-0.091)

(0.048-0.081)

(0.140-0.274)

0.789

0.699

0.776

0.029

0.012

0.033

(0.745-0.832)

(0.640-0.753)

(0.733-0.821)

(0.024-0.035)

(0.006-0.018)

(0.014-0.053)

0.571

0.577

0.570

0.001

0.002

(0.386-0.712)

(0.456-0.683)

(0.386-0.712)

(0.000-0.002)

(0.001-0.002)

(0.002-0.004)

0.596

0.613

0.595

0.002

0.009

(0.390-0.743)

(0.418-0.758)

(0.390-0.740)

(0.001-0.004)

(0.005-0.015)

0.722

0.640

0.719

0.001

0.005

(0.525-0.888)

(0.502-0.713)

(0.523-0.886)

(0.001-0.002)

(0.002-0.008)

0.634

0.601

0.604

0.065

0.045

0.111

(0.543-0.723)

(0.520-0.681)

(0.513-0.689)

(0.047-0.084)

(0.032-0.061)

(0.063-0.217)

0.498

0.514

0.470

0.077

0.081

0.192

(0.381-0.618)

(0.448-0.569)

(0.382-0.576)

(0.053-0.105)

(0.054-0.106)

(0.121-0.276)

0.524

0.487

0.506

0.084

0.100

0.402

(0.441-0.619)

(0.407-0.556)

(0.428-0.600)

(0.066-0.108)

(0.076-0.126)

(0.289-0.519)

0.630

0.588

0.592

0.102

0.056

0.136

(0.550-0.703)

(0.539-0.633)

(0.525-0.662)

(0.083-0.122)

(0.037-0.077)

(0.080-0.197)

0.484

0.487

0.489

0.148

0.109

0.375

(0.379-0.574)

(0.418-0.550)

(0.392-0.566)

(0.120-0.177)

(0.081-0.140)

(0.281-0.474)

0.644

0.612

0.618

0.115

0.083

0.307

(0.568-0.707)

(0.533-0.685)

(0.546-0.678)

(0.095-0.137)

(0.059-0.111)

(0.181-0.422)

0.737

0.670

0.722

0.116

0.061

0.154

(0.657-0.814)

(0.594-0.737)

(0.647-0.795)

(0.094-0.138)

(0.037-0.087)

(0.081-0.245)

0.693

0.614

0.686

0.158

0.114

0.310

(0.581-0.793)

(0.539-0.678)

(0.602-0.771)

(0.124-0.196)

(0.064-0.165)

(0.159-0.508)

0.754

0.676

0.741

0.116

0.068

0.187

(0.693-0.813)

(0.617-0.730)

(0.688-0.796)

(0.095-0.141)

(0.043-0.107)

(0.104-0.302)

0.700

0.623

0.635

0.215

0.092

0.220

(0.632-0.762)

(0.566-0.690)

(0.588-0.680)

(0.196-0.234)

(0.049-0.142)

(0.110-0.377)

0.541

0.579

0.503

0.343

0.320

0.741

(0.444-0.668)

(0.490-0.675)

(0.432-0.583)

(0.282-0.406)

(0.240-0.397)

(0.510-0.983)

0.700

0.662

0.638

0.225

0.137

0.309

(0.634-0.762)

(0.602-0.731)

(0.582-0.686)

(0.198-0.255)

(0.095-0.183)

(0.207-0.446)

0.658

0.614

0.628

0.156

0.045

0.108

(0.617-0.696)

(0.584-0.648)

(0.594-0.663)

(0.145-0.167)

(0.024-0.061)

(0.058-0.169)

0.618

0.587

0.619

0.201

0.115

0.373

(0.559-0.678)

(0.526-0.643)

(0.568-0.670)

(0.173-0.225)

(0.081-0.156)

(0.240-0.491)

0.789

0.723

0.751

0.130

0.035

0.083

(0.753-0.819)

(0.692-0.757)

(0.724-0.776)

(0.118-0.141)

(0.018-0.052)

(0.044-0.137)

0.700

0.602

0.686

0.005

0.004

0.012

(0.608-0.784)

(0.537-0.651)

(0.593-0.769)

(0.003-0.008)

(0.002-0.005)

(0.005-0.022)

0.514

0.533

0.517

0.004

0.011

0.066

(0.136-0.839)

(0.524-0.541)

(0.124-0.846)

(0.002-0.007)

(0.009-0.014)

(0.049-0.086)

0.694

0.668

0.692

0.005

0.004

0.011

(0.570-0.811)

(0.574-0.781)

(0.570-0.808)

(0.003-0.008)

(0.002-0.005)

(0.005-0.020)

0.571

0.534

0.549

0.031

0.010

0.028

(0.520-0.633)

(0.493-0.585)

(0.502-0.612)

(0.026-0.037)

(0.006-0.015)

(0.014-0.049)

0.609

0.578

0.597

0.046

0.026

0.069

(0.521-0.677)

(0.503-0.646)

(0.516-0.668)

(0.036-0.057)

(0.016-0.038)

(0.044-0.118)

0.787

0.681

0.761

0.029

0.011

0.042

(0.740-0.834)

(0.635-0.736)

(0.716-0.809)

(0.024-0.035)

(0.006-0.017)

(0.016-0.077)

0.635

0.614

0.603

0.140

0.050

0.110

(0.588-0.676)

(0.577-0.659)

(0.564-0.647)

(0.123-0.157)

(0.031-0.073)

(0.069-0.169)

Continued on next page

Last Progress A POLLO Breast Cancer Metastasis

Age-Sex Last Progress A POLLO

Breast Cancer Mortality

Age-Sex Last Progress A POLLO

Breast Cancer Palliative Care

Age-Sex Last Progress A POLLO

Chronic Lymphocytic Leukemia Mortality

Age-Sex Last Progress A POLLO

Chronic Myeloid Leukemia Mortality

Age-Sex Last Progress A POLLO

Cirrhosis to Liver Cancer

Age-Sex Last Progress∗∗ A POLLO

Colorectal Cancer Metastasis

Age-Sex Last Progress A POLLO

Colorectal Cancer Mortality

Age-Sex Last Progress A POLLO

GERD to Barrett’s Esophagus

Age-Sex Last Progress A POLLO

Inflammatory Bowel Disease to Colorectal Cancer

Age-Sex Last Progress A POLLO

Kidney Cancer Metastasis

Age-Sex Last Progress

0.641

0.628

0.627

0.159

0.108

0.343

(0.580-0.707)

(0.583-0.681)

(0.572-0.686)

(0.140-0.182)

(0.080-0.147)

(0.225-0.472)

0.762

0.680

0.727

0.125

0.050

0.126

(0.727-0.797)

(0.649-0.712)

(0.697-0.761)

(0.108-0.139)

(0.031-0.071)

(0.069-0.201)

0.598

0.584

0.605

0.036

0.010

0.028

(0.557-0.646)

(0.546-0.613)

(0.566-0.652)

(0.031-0.041)

(0.006-0.014)

(0.013-0.047)

0.637

0.618

0.606

0.052

0.043

0.112

(0.576-0.715)

(0.571-0.680)

(0.548-0.684)

(0.040-0.069)

(0.031-0.055)

(0.073-0.165)

0.752

0.672

0.724

0.034

0.012

0.037

(0.715-0.797)

(0.634-0.715)

(0.690-0.766)

(0.030-0.040)

(0.007-0.016)

(0.021-0.059)

0.632

0.600

0.621

0.057

0.016

0.034

(0.603-0.658)

(0.579-0.628)

(0.594-0.647)

(0.053-0.062)

(0.011-0.021)

(0.021-0.050)

0.724

0.677

0.711

0.062

0.026

0.072

(0.672-0.769)

(0.630-0.715)

(0.664-0.755)

(0.051-0.072)

(0.015-0.039)

(0.032-0.131)

0.826

0.761

0.817

0.048

0.009

0.025

(0.799-0.852)

(0.735-0.782)

(0.793-0.842)

(0.044-0.052)

(0.005-0.013)

(0.012-0.042)

0.488

0.485

0.479

0.019

0.008

0.018

(0.434-0.557)

(0.455-0.529)

(0.427-0.545)

(0.016-0.023)

(0.005-0.011)

(0.011-0.030)

0.608

0.575

0.586

0.034

0.037

0.161

(0.520-0.686)

(0.520-0.637)

(0.501-0.667)

(0.026-0.043)

(0.026-0.048)

(0.103-0.210)

0.765

0.696

0.736

0.019

0.013

0.064

(0.708-0.804)

(0.644-0.741)

(0.680-0.775)

(0.015-0.023)

(0.009-0.017)

(0.032-0.093)

0.662

0.638

0.118

0.061

0.152

(0.578-0.743)

(0.597-0.717)

(0.567-0.710)

(0.099-0.143)

(0.037-0.084)

(0.088-0.264)

0.508

0.429

0.534

0.187

0.194

0.486

(0.362-0.668)

(0.313-0.545)

(0.400-0.659)

(0.122-0.250)

(0.118-0.271)

(0.293-0.850)

0.745

0.675

0.728

0.114

0.078

0.279

(0.666-0.807)

(0.614-0.740)

(0.658-0.785)

(0.091-0.140)

(0.052-0.108)

(0.131-0.441)

0.746

0.675

0.712

0.108

0.075

0.184

(0.660-0.840)

(0.609-0.745)

(0.634-0.801)

(0.089-0.130)

(0.052-0.106)

(0.122-0.247)

0.524

0.626

0.566

0.170

0.197

0.616

(0.321-0.741)

(0.484-0.745)

(0.392-0.756)

(0.109-0.239)

(0.105-0.285)

(0.253-0.990)

0.833

0.770

0.802

0.100

0.055

0.150

(0.755-0.886)

(0.706-0.822)

(0.731-0.855)

(0.076-0.122)

(0.032-0.081)

(0.075-0.270)

0.683

0.613

0.662

0.028

0.020

0.047

(0.593-0.763)

(0.523-0.707)

(0.570-0.744)

(0.019-0.037)

(0.013-0.029)

(0.026-0.083)

0.414

0.481

0.438

0.040

0.052

0.229

(0.249-0.575)

(0.405-0.556)

(0.267-0.592)

(0.026-0.057)

(0.036-0.072)

(0.142-0.319)

0.512

0.523

0.525

0.033

0.037

0.133

(0.399-0.647)

(0.428-0.604)

(0.411-0.658)

(0.025-0.045)

(0.025-0.048)

(0.071-0.179)

0.583

0.550

0.583

0.056

0.021

0.055

(0.528-0.651)

(0.493-0.601)

(0.530-0.649)

(0.045-0.068)

(0.013-0.032)

(0.029-0.088)

0.526

0.491

0.530

0.074

0.067

0.177

(0.433-0.627)

(0.483-0.498)

(0.442-0.617)

(0.054-0.101)

(0.043-0.093)

(0.102-0.249)

0.716

0.640

0.687

0.056

0.037

0.118

(0.663-0.767)

(0.586-0.687)

(0.636-0.740)

(0.045-0.066)

(0.023-0.048)

(0.059-0.185)

0.544

0.542

0.537

0.152

0.045

0.095

(0.512-0.587)

(0.508-0.568)

(0.505-0.578)

(0.139-0.163)

(0.030-0.062)

(0.062-0.134)

0.601

0.594

0.586

0.173

0.117

0.294

(0.540-0.660)

(0.543-0.655)

(0.527-0.644)

(0.150-0.197)

(0.087-0.146)

(0.200-0.383)

0.781

0.713

0.751

0.124

0.033

0.087

(0.748-0.811)

(0.680-0.747)

(0.723-0.777)

(0.114-0.135)

(0.021-0.046)

(0.049-0.131)

0.619

0.608

0.611

0.011

0.003

0.009

(0.592-0.645)

(0.589-0.628)

(0.585-0.638)

(0.009-0.012)

(0.003-0.004)

(0.005-0.013)

0.592

0.588

0.603

0.012

0.007

0.023

(0.534-0.649)

(0.541-0.626)

(0.547-0.662)

(0.010-0.014)

(0.005-0.009)

(0.013-0.032)

0.674

0.626

0.668

0.011

0.004

0.011

(0.652-0.705)

(0.599-0.651)

(0.646-0.698)

(0.009-0.012)

(0.003-0.005)

(0.007-0.016)

0.643

0.588

0.634

0.003

0.008

(0.502-0.769)

(0.491-0.692)

(0.494-0.756)

(0.002-0.005)

(0.002-0.004)

(0.005-0.013)

0.772

0.482

0.759

0.012

0.015

0.136

(0.388-0.929)

(0.475-0.489)

(0.375-0.932)

(0.002-0.019)

(0.004-0.019)

(0.037-0.182)

0.665

0.652

0.663

0.003

0.004

0.026

(0.574-0.755)

(0.572-0.725)

(0.574-0.751)

(0.002-0.005)

(0.003-0.005)

(0.018-0.032)

0.521

0.476

0.517

0.057

0.031

0.072

(0.443-0.597)

(0.417-0.552)

(0.442-0.592)

(0.041-0.075)

(0.020-0.044)

(0.044-0.126)

0.658

0.644

0.645

0.076

0.069

0.203

(0.532-0.784)

(0.551-0.750)

(0.522-0.765)

(0.054-0.103)

(0.043-0.096)

(0.121-0.325)

Continued on next page

A POLLO Kidney Cancer Mortality

Age-Sex Last Progress A POLLO

Liver Cancer Metastasis

Age-Sex Last Progress A POLLO

Liver Cancer Mortality

Age-Sex Last Progress A POLLO

Lung Cancer Metastasis

Age-Sex Last Progress A POLLO

Lung Cancer Mortality

Age-Sex Last Progress A POLLO

Melanoma Metastasis

Age-Sex Last Progress A POLLO

Melanoma Mortality

Age-Sex Last Progress A POLLO

Myelodysplastic Leukemia

Syndrome

Age-Sex Last Progress A POLLO

Ovarian Cancer Metastasis

Age-Sex Last Progress A POLLO

Ovarian Cancer Mortality

Age-Sex Last Progress A POLLO

Pancreatic Cancer Metastasis

Age-Sex Last Progress A POLLO

Pancreatic Cancer Mortality

Age-Sex

0.733

0.651

0.715

0.052

0.032

0.086

(0.646-0.814)

(0.576-0.724)

(0.632-0.795)

(0.039-0.066)

(0.020-0.045)

(0.045-0.139)

0.593

0.546

0.578

0.127

0.056

0.125

(0.527-0.654)

(0.496-0.585)

(0.514-0.635)

(0.110-0.143)

(0.035-0.077)

(0.086-0.191)

0.717

0.662

0.691

0.108

0.067

0.186

(0.629-0.802)

(0.589-0.750)

(0.605-0.774)

(0.080-0.136)

(0.036-0.097)

(0.095-0.302)

0.761

0.665

0.743

0.110

0.055

0.153

(0.711-0.810)

(0.619-0.710)

(0.699-0.789)

(0.095-0.129)

(0.028-0.079)

(0.081-0.262)

0.551

0.536

0.565

0.034

0.031

0.062

(0.443-0.677)

(0.460-0.618)

(0.465-0.687)

(0.024-0.044)

(0.022-0.040)

(0.044-0.109)

0.573

0.491

0.557

0.043

0.041

0.114

(0.426-0.720)

(0.374-0.592)

(0.412-0.704)

(0.026-0.059)

(0.021-0.060)

(0.041-0.215)

0.693

0.649

0.676

0.033

0.024

0.068

(0.594-0.805)

(0.564-0.745)

(0.578-0.785)

(0.022-0.043)

(0.015-0.033)

(0.038-0.093)

0.552

0.545

0.554

0.210

0.084

0.204

(0.489-0.606)

(0.500-0.593)

(0.503-0.599)

(0.190-0.236)

(0.052-0.125)

(0.112-0.316)

0.584

0.554

0.569

0.276

0.224

0.442

(0.476-0.675)

(0.459-0.632)

(0.489-0.637)

(0.237-0.322)

(0.167-0.283)

(0.311-0.634)

0.638

0.625

0.218

0.141

0.302

(0.576-0.698)

(0.570-0.666)

(0.571-0.673)

(0.188-0.249)

(0.096-0.191)

(0.204-0.446)

0.559

0.537

0.562

0.075

0.027

0.066

(0.509-0.601)

(0.491-0.577)

(0.516-0.601)

(0.065-0.087)

(0.017-0.040)

(0.039-0.098)

0.612

0.589

0.593

0.106

0.075

0.211

(0.549-0.664)

(0.533-0.640)

(0.534-0.643)

(0.089-0.121)

(0.055-0.092)

(0.129-0.293)

0.725

0.685

0.703

0.071

0.025

0.073

(0.688-0.767)

(0.646-0.723)

(0.666-0.744)

(0.061-0.082)

(0.014-0.035)

(0.031-0.127)

0.585

0.577

0.567

0.212

0.046

0.106

(0.562-0.610)

(0.556-0.596)

(0.545-0.590)

(0.205-0.220)

(0.031-0.062)

(0.059-0.171)

0.688

0.639

0.660

0.198

0.056

0.141

(0.658-0.722)

(0.610-0.667)

(0.634-0.688)

(0.187-0.212)

(0.038-0.080)

(0.092-0.228)

0.727

0.660

0.691

0.186

0.052

0.123

(0.706-0.750)

(0.640-0.684)

(0.673-0.711)

(0.177-0.196)

(0.037-0.072)

(0.076-0.190)

0.596

0.574

0.587

0.030

0.015

0.040

(0.518-0.658)

(0.517-0.634)

(0.508-0.651)

(0.021-0.039)

(0.009-0.022)

(0.019-0.069)

0.543

0.524

0.542

0.062

0.068

0.197

(0.448-0.639)

(0.433-0.606)

(0.444-0.633)

(0.047-0.081)

(0.050-0.091)

(0.122-0.294)

0.730

0.641

0.714

0.030

0.022

0.080

(0.649-0.800)

(0.579-0.705)

(0.631-0.784)

(0.024-0.038)

(0.014-0.031)

(0.037-0.128)

0.705

0.656

0.683

0.066

0.024

0.059

(0.666-0.741)

(0.621-0.686)

(0.644-0.716)

(0.058-0.076)

(0.015-0.037)

(0.030-0.094)

0.666

0.651

0.664

0.085

0.046

0.117

(0.582-0.758)

(0.592-0.713)

(0.584-0.746)

(0.065-0.106)

(0.029-0.064)

(0.068-0.191)

0.869

0.765

0.848

0.057

0.015

0.052

(0.837-0.893)

(0.732-0.793)

(0.815-0.872)

(0.048-0.065)

(0.008-0.024)

(0.021-0.097)

0.510

0.486

0.524

0.073

0.038

0.090

(0.445-0.588)

(0.426-0.534)

(0.465-0.597)

(0.058-0.087)

(0.025-0.051)

(0.051-0.154)

0.612

0.549

0.607

0.098

0.066

0.191

(0.521-0.704)

(0.468-0.621)

(0.515-0.690)

(0.076-0.121)

(0.042-0.095)

(0.095-0.327)

0.683

0.570

0.673

0.069

0.035

0.079

(0.598-0.756)

(0.526-0.610)

(0.590-0.747)

(0.057-0.085)

(0.020-0.046)

(0.050-0.116)

0.592

0.557

0.563

0.040

0.025

0.059

(0.512-0.653)

(0.500-0.606)

(0.485-0.621)

(0.031-0.049)

(0.015-0.035)

(0.035-0.090)

0.576

0.584

0.558

0.064

0.047

0.119

(0.484-0.665)

(0.505-0.654)

(0.464-0.652)

(0.046-0.084)

(0.030-0.068)

(0.064-0.169)

0.759

0.661

0.725

0.039

0.027

0.085

(0.690-0.819)

(0.606-0.723)

(0.660-0.791)

(0.030-0.049)

(0.018-0.039)

(0.046-0.154)

0.655

0.594

0.618

0.203

0.097

0.233

(0.599-0.704)

(0.545-0.634)

(0.575-0.655)

(0.182-0.224)

(0.069-0.129)

(0.141-0.334)

0.629

0.548

0.590

0.256

0.212

0.469

(0.543-0.711)

(0.490-0.616)

(0.524-0.658)

(0.211-0.298)

(0.154-0.268)

(0.333-0.669)

0.753

0.667

0.715

0.182

0.095

0.227

(0.699-0.798)

(0.624-0.706)

(0.671-0.754)

(0.158-0.202)

(0.063-0.123)

(0.127-0.352)

0.506

0.457

0.517

0.065

0.034

0.087

(0.431-0.575)

(0.427-0.493)

(0.451-0.579)

(0.051-0.075)

(0.024-0.047)

(0.054-0.145)

0.509

0.451

0.499

0.081

0.053

0.136

(0.436-0.592)

(0.398-0.513)

(0.431-0.588)

(0.059-0.104)

(0.031-0.076)

(0.074-0.217)

0.639

0.552

0.626

0.064

0.034

0.094

(0.579-0.693)

(0.516-0.587)

(0.566-0.680)

(0.053-0.079)

(0.022-0.047)

(0.052-0.153)

0.563

0.546

0.552

0.242

0.077

0.183

(0.520-0.605)

(0.513-0.582)

(0.524-0.576)

(0.236-0.248)

(0.051-0.107)

(0.104-0.289)

Continued on next page

Last Progress A POLLO Prostate Cancer Metastasis

Age-Sex Last Progress A POLLO

Prostate Cancer Mortality

Age-Sex Last Progress A POLLO

Thyroid Cancer Metastasis

Age-Sex Last Progress A POLLO

Thyroid Cancer Mortality

Age-Sex Last Progress∗ A POLLO

Acute COVID-19 to Long COVID

Age-Sex Last Progress A POLLO

Asthma to Exacerbation Requiring Oral Corticosteroids

Age-Sex Last Progress A POLLO

COPD to Chronic Respiratory Failure

Age-Sex Last Progress A POLLO

COPD to Exacerbation

Age-Sex Last Progress A POLLO

COPD to Pulmonary Hypertension

Age-Sex Last Progress A POLLO

COVID-19 Mortality

60d

Age-Sex Last Progress A POLLO

Pulmonary Fibrosis to Pulmonary Hypertension

Age-Sex Last Progress A POLLO

0.571

0.566

0.548

0.270

0.167

0.358

(0.494-0.643)

(0.500-0.623)

(0.510-0.585)

(0.247-0.296)

(0.116-0.220)

(0.250-0.493)

0.649

0.613

0.608

0.236

0.101

0.244

(0.607-0.703)

(0.570-0.658)

(0.584-0.642)

(0.220-0.250)

(0.068-0.133)

(0.159-0.365)

0.534

0.511

0.516

0.029

0.010

0.023

(0.484-0.603)

(0.489-0.550)

(0.465-0.579)

(0.024-0.034)

(0.006-0.015)

(0.012-0.035)

0.678

0.610

0.676

0.040

0.025

0.087

(0.580-0.769)

(0.523-0.679)

(0.586-0.767)

(0.027-0.051)

(0.014-0.036)

(0.029-0.156)

0.705

0.672

0.673

0.029

0.021

0.097

(0.659-0.755)

(0.626-0.722)

(0.620-0.721)

(0.023-0.034)

(0.017-0.027)

(0.062-0.136)

0.743

0.678

0.721

0.095

0.021

0.052

(0.723-0.761)

(0.660-0.702)

(0.704-0.739)

(0.087-0.103)

(0.014-0.029)

(0.031-0.077)

0.730

0.674

0.732

0.111

0.033

0.081

(0.679-0.776)

(0.640-0.708)

(0.693-0.770)

(0.096-0.130)

(0.020-0.052)

(0.050-0.130)

0.839

0.775

0.814

0.082

0.019

0.060

(0.822-0.860)

(0.752-0.795)

(0.797-0.835)

(0.074-0.089)

(0.012-0.028)

(0.029-0.104)

0.649

0.612

0.636

0.017

0.008

0.023

(0.566-0.723)

(0.554-0.672)

(0.553-0.707)

(0.013-0.021)

(0.005-0.011)

(0.009-0.041)

0.606

0.587

0.620

0.024

0.019

0.053

(0.489-0.742)

(0.487-0.686)

(0.504-0.753)

(0.015-0.035)

(0.010-0.029)

(0.027-0.108)

0.774

0.692

0.753

0.017

0.009

0.032

(0.712-0.834)

(0.641-0.747)

(0.689-0.816)

(0.013-0.021)

(0.005-0.014)

(0.014-0.062)

0.806

0.686

0.788

0.043

0.022

0.072

(0.756-0.852)

(0.618-0.755)

(0.738-0.839)

(0.034-0.055)

(0.012-0.033)

(0.033-0.125)

0.642

0.694

0.630

0.048

0.044

0.149

(0.466-0.788)

(0.576-0.781)

(0.453-0.768)

(0.032-0.067)

(0.025-0.066)

(0.071-0.263)

0.828

0.741

0.814

0.044

0.028

0.132

(0.765-0.898)

(0.672-0.818)

(0.751-0.880)

(0.034-0.056)

(0.015-0.039)

(0.042-0.217)

0.561

0.581

0.554

0.005

0.003

0.007

(0.485-0.622)

(0.527-0.630)

(0.478-0.614)

(0.004-0.006)

(0.002-0.004)

(0.004-0.012)

0.491

0.479

0.495

0.006

0.021

(0.371-0.563)

(0.387-0.550)

(0.377-0.568)

(0.005-0.008)

(0.012-0.029)

0.644

0.569

0.646

0.005

0.015

(0.553-0.732)

(0.507-0.620)

(0.555-0.733)

(0.004-0.007)

(0.003-0.007)

(0.008-0.024)

0.574

0.588

0.570

0.024

0.007

0.022

(0.547-0.605)

(0.567-0.610)

(0.543-0.601)

(0.022-0.026)

(0.005-0.009)

(0.014-0.030)

0.628

0.607

0.623

0.056

0.010

0.026

(0.600-0.662)

(0.582-0.633)

(0.595-0.656)

(0.050-0.061)

(0.006-0.015)

(0.014-0.044)

0.830

0.754

0.810

0.023

0.005

0.014

(0.815-0.850)

(0.736-0.774)

(0.795-0.831)

(0.021-0.024)

(0.004-0.008)

(0.007-0.028)

0.627

0.601

0.604

0.013

0.007

0.016

(0.566-0.695)

(0.551-0.643)

(0.541-0.673)

(0.010-0.017)

(0.004-0.009)

(0.009-0.023)

0.549

0.504

0.543

0.023

0.024

0.104

(0.438-0.650)

(0.444-0.558)

(0.433-0.645)

(0.015-0.031)

(0.018-0.031)

(0.065-0.132)

0.723

0.647

0.704

0.013

0.009

0.027

(0.643-0.783)

(0.563-0.722)

(0.625-0.766)

(0.010-0.016)

(0.006-0.012)

(0.013-0.050)

0.572

0.553

0.561

0.023

0.008

0.018

(0.529-0.625)

(0.512-0.597)

(0.519-0.613)

(0.019-0.027)

(0.005-0.011)

(0.011-0.028)

0.537

0.532

0.539

0.049

0.032

0.110

(0.464-0.604)

(0.488-0.573)

(0.473-0.604)

(0.038-0.060)

(0.024-0.043)

(0.064-0.145)

0.791

0.738

0.780

0.023

0.009

0.035

(0.751-0.833)

(0.702-0.778)

(0.739-0.823)

(0.019-0.026)

(0.006-0.013)

(0.015-0.055)

0.654

0.612

0.636

0.017

0.008

0.024

(0.605-0.702)

(0.565-0.657)

(0.587-0.682)

(0.013-0.020)

(0.005-0.011)

(0.014-0.041)

0.572

0.528

0.555

0.029

0.020

0.054

(0.478-0.659)

(0.445-0.610)

(0.464-0.644)

(0.022-0.037)

(0.013-0.028)

(0.033-0.085)

0.723

0.663

0.711

0.017

0.008

0.025

(0.667-0.772)

(0.604-0.706)

(0.654-0.760)

(0.014-0.020)

(0.005-0.011)

(0.013-0.041)

0.817

0.752

0.817

0.002

0.001

0.004

(0.736-0.894)

(0.645-0.870)

(0.736-0.894)

(0.001-0.003)

(0.001-0.002)

(0.002-0.008)

0.845

0.680

0.844

0.002

0.001

0.004

(0.717-0.944)

(0.567-0.817)

(0.717-0.943)

(0.001-0.003)

(0.000-0.002)

(0.001-0.010)

0.869

0.551

0.868

0.001

0.000

0.003

(0.742-0.960)

(0.496-0.697)

(0.741-0.960)

(0.000-0.001)

(0.001-0.007)

0.581

0.564

0.552

0.055

0.029

0.076

(0.506-0.664)

(0.509-0.615)

(0.481-0.627)

(0.042-0.068)

(0.018-0.041)

(0.041-0.125)

0.496

0.479

0.506

0.061

0.055

0.154

(0.396-0.605)

(0.406-0.547)

(0.403-0.606)

(0.048-0.080)

(0.039-0.074)

(0.100-0.218)

0.635

0.589

0.637

0.054

0.028

0.067

(0.552-0.719)

(0.533-0.664)

(0.557-0.719)

(0.043-0.069)

(0.016-0.039)

(0.036-0.109)

Extended Data Table 10: Definition of treatment response prediction tasks. We evaluated TTE performance on 59 treatment response tasks. τ is the threshold duration. Incidence refers to the percentage of patients in the test set who reach the endpoint by the specified τ . The Snapshot Event column specifies all medications used to identify the treatment indicated in the Task column. Integers refer to ingredient RxCUIs, and the ∧ symbol indicates combination therapies where both ingredients must be present. The Endpoint Event column specifies all ICD-10 codes used to identify the disease indicated in the Task column. Category

Task

Incid.

Snapshot Event

Endpoint Event

Cardiovascular & Circulatory

MACE-Free Survival on Amlodipine for Hypertension MACE-Free Survival on Aspirin for Chronic Ischemic Heart Disease MACE-Free Survival on Atorvastatin for Chronic Ischemic Heart Disease MACE-Free Survival on Chlorthalidone for Hypertension MACE-Free Survival on Clopidogrel for Chronic Ischemic Heart Disease MACE-Free Survival on Lisinopril for Hypertension MACE-Free Survival on Losartan for Hypertension MACE-Free Survival on Ticagrelor for Chronic Ischemic Heart Disease Stroke/Embolism-Free Survival on Apixaban for Atrial Fibrillation Stroke/Embolism-Free Survival on Rivaroxaban for Atrial Fibrillation Stroke/Embolism-Free Survival on Warfarin for Atrial Fibrillation Survival on Furosemide for Heart Failure Survival on Lisinopril for Heart Failure Survival on Metoprolol Succinate for Heart Failure Survival on Sacubitril / Valsartan for Heart Failure Survival on Sildenafil for Pulmonary Arterial Hypertension Survival on Spironolactone for Heart Failure HF-Free Survival on Empagliflozin for Type 2 Diabetes ESRD-Free Survival on Lisinopril for Chronic Kidney Disease Survival on Lactulose for Hepatic Encephalopathy Survival on Riluzole for Amyotrophic Lateral Sclerosis Survival on Abiraterone for Prostate Cancer Survival on Atezolizumab for Lung Cancer Survival on Azacitidine for Acute Myeloid Leukemia Survival on Azacitidine for Myelodysplastic Syndrome Survival on Bortezomib for Multiple Myeloma

15,192

26.96%

17767

10,482

27.64%

1191

I21, I22, I60, I61, I62, I63, death I21, I22, I60, I61, I62, I63, death

7,681

31.51%

83367

I21, I22, I60, I61, I62, I63, death

973

22.45%

2409

I21, I22, I60, I61, I62, I63, death

3,019

30.49%

32968

I21, I22, I60, I61, I62, I63, death

14,971

21.36%

29046

8,096

22.37%

52175

2,702

21.71%

1116632

I21, I22, I60, I61, I62, I63, death I21, I22, I60, I61, I62, I63, death I21, I22, I60, I61, I62, I63, death

6,027

35.87%

1364430

G45, I60, I61, I62, I63, I26, I74, I82, death

1,705

26.55%

1114195

G45, I60, I61, I62, I63, I26, I74, I82, death

2,687

35.26%

11289

G45, I60, I61, I62, I63, I26, I74, I82, death

8,343

30.01%

4603

death

2,944

21.85%

29046

death

7,288

28.71%

6918

death

942

28.85%

death

1,056

46.34%

69749∧1656328, 1656328∧69749, 1656339 136411

2,681

31.36%

9997

death

3,990

15.47%

1545653

I50, death

3,085

30.39%

29046

N18.6, death

1,195

40.54%

6218

death

401

60.34%

35623

512

48.10%

1100072

Z43.0, Z93.0, Z99.1, death death

833

52.21%

1792776

death

540

69.14%

1251

death

464

62.16%

1251

death

511

20.55%

358258

death

Cardiovascular & Circulatory

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory

Cardiovascular & Circulatory

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Endocrine, Metabolic & Reproductive Gastrointestinal, Hepatic & Renal

Gastrointestinal, Hepatic & Renal Neurology & Psychiatry Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers)

death

J95.0,

Continued on next page

Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers)

Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Respiratory & Infectious Respiratory & Infectious

Survival on Carboplatin for Endometrial Cancer Survival on Carboplatin for Esophageal Cancer Survival on Carboplatin for Lung Cancer Survival on Carboplatin for Ovarian Cancer Survival on Cisplatin for Bladder Cancer Survival on Cisplatin for Head & Neck Cancer Survival on Cytarabine for Acute Myeloid Leukemia Survival on Doxorubicin for Soft Tissue Sarcoma Survival on Enfortumab vedotin for Bladder Cancer Survival on Fluorouracil for Colorectal Cancer Survival on Gemcitabine for Pancreatic Cancer Survival on Ipilimumab for Kidney Cancer Survival on Leuprolide for Prostate Cancer Survival on Nivolumab for Kidney Cancer Survival on Nivolumab for Melanoma Survival on Octreotide for Neuroendocrine Tumors Survival on Osimertinib for Lung Cancer Survival on Oxaliplatin for Stomach Cancer Survival on Paclitaxel for Breast Cancer Survival on Paclitaxel for Ovarian Cancer Survival on Paclitaxel for Pancreatic Cancer Survival on Pembrolizumab for Breast Cancer Survival on Pembrolizumab for Colorectal Cancer Survival on Pembrolizumab for Lung Cancer Survival on Pembrolizumab for Melanoma Survival on Pemetrexed for Mesothelioma Survival on Pertuzumab for Breast Cancer Survival on Rituximab for Diffuse Large B-Cell Lymphoma Survival on Rituximab for Non-Hodgkin Lymphoma Survival on Temozolomide for Glioblastoma Survival on Trastuzumab for Breast Cancer Exacerbation-Free Survival on Tiotropium for COPD Exacerbation-Free Survival on Umeclidinium / Vilanterol for COPD

1,651

29.60%

40048

death

1,011

39.44%

40048

death

1,339

40.31%

40048

death

674

27.91%

40048

death

1,075

28.79%

2555

death

2,479

13.73%

2555

death

1,003

43.64%

3041

death

1,007

50.36%

3639

death

405

77.42%

2268307

death

1,081

35.47%

4492

death

571

56.77%

12574

death

370

50.82%

1094833

death

2,271

12.26%

42375

death

721

52.43%

1597876

death

1,463

29.49%

1597876

death

806

21.65%

7617

death

393

47.46%

1721560

death

1,129

62.50%

32592

death

1,439

21.43%

56946

death

620

33.14%

56946

death

483

62.88%

56946

death

1,223

61.90%

1547545

death

402

54.55%

1547545

death

756

45.33%

1547545

death

1,271

23.33%

1547545

death

362

58.93%

68446

death

1,464

7.12%

1298944

death

640

21.88%

121191

death

1,225

30.23%

121191

death

498

46.59%

37776

death

4,183

14.77%

224905

death

1,625

45.53%

69120

J44.0, J44.1, death

1,160

53.10%

1487518, 1487514∧1424884, 1424884∧1487514

J44.0, J44.1, death

Extended Data Table 11: Evaluation results of treatment response prediction tasks. We report timedependent performance metrics computed at τ on the test set for 59 treatment response tasks. τ is the threshold duration. Arrows indicate whether higher (↑) or lower (↓) values are better. ICI, Integrated Calibration Index; MCE, Maximum Calibration Error. 10−4 is the default value of the penalizer (λ) for Cox proportional hazards; if convergence cannot be achieved then λ is increased to 1e-3 (∗ ) or 1e-2 (∗∗ ). Mean and 95% CI reported over 100 bootstraps of the test set. Task

Model

MACE-Free Survival on Amlodipine for Hypertension

Age-Sex Last Progress A POLLO

MACE-Free Survival on Aspirin for Chronic Ischemic Heart Disease

Age-Sex Last Progress A POLLO

MACE-Free Survival on Atorvastatin for Chronic Ischemic Heart Disease

Age-Sex

Last Progress A POLLO MACE-Free Chlorthalidone sion

Survival on for Hyperten-

Age-Sex

Last Progress A POLLO MACE-Free Survival on Clopidogrel for Chronic Ischemic Heart Disease

Age-Sex

Last Progress A POLLO MACE-Free Survival on Lisinopril for Hypertension

Age-Sex Last Progress A POLLO

MACE-Free Survival on Losartan for Hypertension

Age-Sex Last Progress A POLLO

MACE-Free Survival on Ticagrelor for Chronic Ischemic Heart Disease

Age-Sex Last Progress A POLLO

Stroke/Embolism-Free Survival on Apixaban for Atrial Fibrillation

Age-Sex Last Progress

AUC (↑)

Bal. Acc. (↑)

C-Index (↑)

Brier Score (↓)

ICI (↓)

MCE (↓)

0.634

0.585

0.608

0.156

0.026

0.055

(0.615-0.655)

(0.568-0.605)

(0.591-0.624)

(0.150-0.163)

(0.015-0.036)

(0.030-0.079)

0.674

0.624

0.657

0.152

0.026

0.065

(0.653-0.695)

(0.603-0.641)

(0.638-0.676)

(0.145-0.159)

(0.016-0.035)

(0.038-0.100)

0.792

0.732

0.765

0.132

0.022

0.069

(0.773-0.810)

(0.713-0.749)

(0.749-0.782)

(0.124-0.138)

(0.012-0.029)

(0.029-0.122)

0.636

0.595

0.618

0.166

0.032

0.076

(0.614-0.662)

(0.575-0.614)

(0.597-0.642)

(0.156-0.173)

(0.020-0.044)

(0.045-0.116)

0.667

0.626

0.650

0.167

0.034

0.084

(0.642-0.689)

(0.604-0.645)

(0.631-0.669)

(0.157-0.178)

(0.023-0.044)

(0.051-0.128)

0.768

0.713

0.747

0.146

0.028

0.071

(0.741-0.790)

(0.691-0.734)

(0.725-0.765)

(0.137-0.155)

(0.015-0.041)

(0.037-0.125)

0.624

0.590

0.601

0.182

0.051

0.133

(0.598-0.654)

(0.564-0.613)

(0.577-0.625)

(0.172-0.191)

(0.040-0.067)

(0.095-0.178)

0.659

0.611

0.646

0.179

0.047

0.109

(0.631-0.689)

(0.588-0.633)

(0.624-0.668)

(0.168-0.189)

(0.032-0.065)

(0.073-0.153)

0.753

0.676

0.723

0.158

0.042

0.103

(0.728-0.776)

(0.653-0.699)

(0.704-0.742)

(0.149-0.167)

(0.027-0.057)

(0.064-0.160)

0.641

0.594

0.616

0.137

0.084

0.245

(0.558-0.746)

(0.519-0.679)

(0.540-0.700)

(0.107-0.178)

(0.054-0.129)

(0.113-0.441)

0.613

0.575

0.616

0.167

0.121

0.303

(0.504-0.705)

(0.489-0.656)

(0.521-0.700)

(0.127-0.208)

(0.072-0.158)

(0.142-0.460)

0.697

0.670

0.691

0.148

0.097

0.263

(0.594-0.795)

(0.601-0.737)

(0.595-0.786)

(0.114-0.184)

(0.058-0.133)

(0.137-0.391)

0.618

0.602

0.597

0.177

0.060

0.150

(0.567-0.669)

(0.559-0.646)

(0.558-0.633)

(0.161-0.192)

(0.035-0.085)

(0.080-0.233)

0.670

0.591

0.660

0.175

0.056

0.154

(0.625-0.713)

(0.541-0.631)

(0.624-0.701)

(0.159-0.193)

(0.035-0.083)

(0.087-0.286)

0.728

0.684

0.708

0.167

0.063

0.159

(0.675-0.767)

(0.639-0.721)

(0.662-0.743)

(0.150-0.185)

(0.040-0.090)

(0.087-0.260)

0.638

0.602

0.619

0.135

0.024

0.053

(0.616-0.661)

(0.582-0.625)

(0.600-0.641)

(0.127-0.143)

(0.013-0.033)

(0.033-0.071)

0.660

0.611

0.644

0.136

0.022

0.054

(0.636-0.683)

(0.589-0.631)

(0.622-0.666)

(0.129-0.143)

(0.012-0.031)

(0.031-0.083)

0.773

0.711

0.755

0.119

0.024

0.082

(0.752-0.789)

(0.692-0.728)

(0.737-0.770)

(0.112-0.127)

(0.015-0.034)

(0.044-0.136)

0.666

0.627

0.646

0.130

0.041

0.094

(0.625-0.701)

(0.597-0.655)

(0.611-0.678)

(0.122-0.139)

(0.032-0.052)

(0.063-0.126)

0.652

0.620

0.639

0.136

0.044

0.157

(0.625-0.687)

(0.594-0.647)

(0.613-0.668)

(0.126-0.147)

(0.031-0.057)

(0.099-0.227)

0.795

0.725

0.777

0.116

0.030

0.103

(0.769-0.819)

(0.698-0.749)

(0.754-0.798)

(0.106-0.126)

(0.021-0.040)

(0.054-0.166)

0.590

0.512

0.576

0.142

0.060

0.146

(0.506-0.666)

(0.472-0.554)

(0.501-0.642)

(0.119-0.164)

(0.040-0.085)

(0.090-0.223)

0.711

0.647

0.671

0.139

0.086

0.224

(0.645-0.777)

(0.579-0.716)

(0.613-0.726)

(0.116-0.162)

(0.056-0.121)

(0.119-0.365)

0.739

0.666

0.710

0.123

0.064

0.172

(0.677-0.813)

(0.614-0.723)

(0.659-0.774)

(0.094-0.146)

(0.040-0.088)

(0.089-0.270)

0.693

0.630

0.655

0.172

0.046

0.126

(0.663-0.729)

(0.604-0.661)

(0.629-0.683)

(0.162-0.183)

(0.028-0.067)

(0.071-0.203)

0.706

0.669

0.686

0.181

0.056

0.161

(0.682-0.736)

(0.641-0.699)

(0.667-0.711)

(0.168-0.191)

(0.035-0.079)

(0.099-0.244)

Continued on next page

A POLLO Stroke/Embolism-Free Survival on Rivaroxaban for Atrial Fibrillation

Age-Sex Last Progress A POLLO

Stroke/Embolism-Free Survival on Warfarin for Atrial Fibrillation

Age-Sex Last Progress A POLLO

Survival on Furosemide for Heart Failure

Age-Sex Last Progress A POLLO

Survival on Lisinopril for Heart Failure

Age-Sex Last Progress A POLLO

Survival on Metoprolol Succinate for Heart Failure

Age-Sex Last Progress A POLLO

Survival on Sacubitril / Valsartan for Heart Failure

Age-Sex Last Progress A POLLO

Survival on Sildenafil for Pulmonary Arterial Hypertension

Age-Sex Last Progress A POLLO

Survival on Spironolactone for Heart Failure

Age-Sex Last Progress A POLLO

HF-Free Survival on Empagliflozin for Type 2 Diabetes

Age-Sex Last Progress A POLLO

ESRD-Free Survival on Lisinopril for Chronic Kidney Disease

Age-Sex Last Progress A POLLO

Survival on Lactulose for Hepatic Encephalopathy

Age-Sex

0.811

0.761

0.785

0.147

0.044

0.113

(0.785-0.841)

(0.736-0.787)

(0.762-0.809)

(0.137-0.160)

(0.029-0.062)

(0.064-0.186)

0.671

0.638

0.641

0.158

0.071

0.174

(0.609-0.729)

(0.586-0.702)

(0.593-0.687)

(0.135-0.178)

(0.038-0.105)

(0.094-0.298)

0.701

0.654

0.671

0.169

0.094

0.236

(0.638-0.748)

(0.596-0.699)

(0.617-0.711)

(0.137-0.195)

(0.062-0.127)

(0.140-0.370)

0.769

0.701

0.733

0.149

0.087

0.216

(0.727-0.810)

(0.659-0.748)

(0.695-0.773)

(0.122-0.171)

(0.058-0.119)

(0.130-0.328)

0.607

0.575

0.594

0.211

0.070

0.167

(0.565-0.646)

(0.541-0.612)

(0.566-0.627)

(0.200-0.223)

(0.041-0.097)

(0.089-0.271)

0.632

0.602

0.615

0.211

0.073

0.174

(0.583-0.678)

(0.558-0.644)

(0.579-0.647)

(0.195-0.228)

(0.049-0.098)

(0.108-0.243)

0.738

0.682

0.708

0.186

0.058

0.146

(0.703-0.769)

(0.645-0.715)

(0.683-0.733)

(0.173-0.202)

(0.037-0.081)

(0.075-0.229)

0.687

0.635

0.658

0.177

0.044

0.109

(0.664-0.712)

(0.615-0.659)

(0.636-0.682)

(0.169-0.186)

(0.027-0.057)

(0.059-0.159)

0.608

0.565

0.593

0.191

0.039

0.106

(0.586-0.631)

(0.541-0.584)

(0.573-0.611)

(0.182-0.202)

(0.024-0.055)

(0.055-0.165)

0.748

0.688

0.720

0.165

0.031

0.074

(0.728-0.769)

(0.670-0.708)

(0.703-0.736)

(0.156-0.172)

(0.019-0.042)

(0.044-0.116)

0.693

0.643

0.667

0.147

0.056

0.151

(0.641-0.741)

(0.601-0.685)

(0.622-0.706)

(0.132-0.159)

(0.031-0.080)

(0.076-0.260)

0.585

0.560

0.576

0.157

0.053

0.121

(0.521-0.637)

(0.505-0.598)

(0.525-0.621)

(0.143-0.176)

(0.033-0.074)

(0.070-0.192)

0.771

0.689

0.744

0.134

0.044

0.114

(0.727-0.805)

(0.648-0.724)

(0.707-0.775)

(0.121-0.147)

(0.028-0.061)

(0.066-0.181)

0.642

0.610

0.619

0.175

0.045

0.098

(0.610-0.671)

(0.587-0.637)

(0.594-0.644)

(0.162-0.184)

(0.029-0.056)

(0.059-0.166)

0.637

0.610

0.623

0.180

0.039

0.119

(0.609-0.667)

(0.584-0.631)

(0.600-0.647)

(0.171-0.192)

(0.025-0.051)

(0.060-0.189)

0.749

0.689

0.720

0.159

0.033

0.080

(0.723-0.772)

(0.667-0.713)

(0.699-0.742)

(0.149-0.170)

(0.019-0.046)

(0.048-0.124)

0.586

0.570

0.590

0.176

0.102

0.256

(0.487-0.674)

(0.498-0.635)

(0.504-0.666)

(0.142-0.215)

(0.060-0.144)

(0.130-0.402)

0.565

0.557

0.542

0.204

0.206

0.520

(0.462-0.652)

(0.466-0.638)

(0.452-0.611)

(0.157-0.265)

(0.163-0.266)

(0.344-0.700)

0.728

0.650

0.716

0.165

0.130

0.323

(0.653-0.809)

(0.592-0.715)

(0.655-0.796)

(0.124-0.203)

(0.086-0.183)

(0.199-0.458)

0.669

0.573

0.593

0.214

0.132

0.315

(0.566-0.751)

(0.475-0.659)

(0.518-0.654)

(0.189-0.243)

(0.092-0.209)

(0.176-0.481)

0.622

0.552

0.575

0.231

0.175

0.404

(0.538-0.718)

(0.479-0.620)

(0.516-0.648)

(0.205-0.253)

(0.124-0.235)

(0.268-0.583)

0.743

0.679

0.668

0.201

0.157

0.384

(0.655-0.826)

(0.596-0.758)

(0.601-0.732)

(0.157-0.242)

(0.111-0.213)

(0.236-0.572)

0.623

0.589

0.594

0.186

0.071

0.182

(0.579-0.677)

(0.547-0.629)

(0.554-0.637)

(0.167-0.200)

(0.050-0.097)

(0.111-0.273)

0.596

0.572

0.579

0.190

0.070

0.165

(0.548-0.646)

(0.533-0.615)

(0.542-0.615)

(0.172-0.208)

(0.045-0.094)

(0.105-0.239)

0.710

0.652

0.681

0.174

0.062

0.146

(0.666-0.749)

(0.619-0.678)

(0.646-0.717)

(0.155-0.196)

(0.040-0.089)

(0.082-0.233)

0.574

0.581

0.561

0.112

0.055

0.121

(0.516-0.634)

(0.531-0.637)

(0.507-0.614)

(0.097-0.129)

(0.037-0.070)

(0.074-0.169)

0.732

0.691

0.714

0.103

0.050

0.124

(0.671-0.782)

(0.644-0.739)

(0.661-0.760)

(0.086-0.118)

(0.032-0.069)

(0.073-0.197)

0.799

0.674

0.780

0.096

0.040

0.104

(0.754-0.844)

(0.624-0.737)

(0.737-0.821)

(0.084-0.109)

(0.024-0.059)

(0.051-0.180)

0.618

0.561

0.593

0.179

0.057

0.134

(0.578-0.672)

(0.530-0.600)

(0.561-0.633)

(0.164-0.194)

(0.034-0.079)

(0.080-0.208)

0.644

0.599

0.632

0.179

0.064

0.160

(0.602-0.689)

(0.556-0.641)

(0.601-0.671)

(0.163-0.195)

(0.038-0.087)

(0.090-0.272)

0.742

0.679

0.716

0.162

0.062

0.150

(0.702-0.785)

(0.643-0.719)

(0.683-0.746)

(0.144-0.180)

(0.038-0.086)

(0.090-0.234)

0.627

0.559

0.578

0.210

0.114

0.264

(0.515-0.729)

(0.501-0.628)

(0.500-0.648)

(0.184-0.240)

(0.077-0.182)

(0.151-0.468)

Continued on next page

Last Progress A POLLO Survival on Riluzole for Amyotrophic Lateral Sclerosis

Age-Sex Last Progress A POLLO

Survival on Abiraterone Prostate Cancer

for

Age-Sex Last Progress A POLLO

Survival on Atezolizumab for Lung Cancer

Age-Sex Last Progress A POLLO

Survival on Azacitidine for Acute Myeloid Leukemia

Age-Sex Last Progress A POLLO

Survival on Azacitidine Myelodysplastic Syndrome

for

Age-Sex Last Progress A POLLO

Survival on Bortezomib for Multiple Myeloma

Age-Sex Last Progress A POLLO

Survival on Carboplatin for Endometrial Cancer

Age-Sex Last Progress A POLLO

Survival on Carboplatin Esophageal Cancer

for

Age-Sex Last Progress A POLLO

Survival on Carboplatin for Lung Cancer

Age-Sex Last Progress A POLLO

Survival on Carboplatin for Ovarian Cancer

Age-Sex Last Progress A POLLO

0.605

0.554

0.567

0.222

0.151

0.387

(0.521-0.690)

(0.483-0.641)

(0.499-0.636)

(0.195-0.248)

(0.089-0.224)

(0.240-0.630)

0.610

0.532

0.586

0.217

0.135

0.324

(0.512-0.714)

(0.466-0.594)

(0.519-0.661)

(0.180-0.252)

(0.084-0.187)

(0.189-0.511)

0.730

0.661

0.646

0.241

0.257

0.534

(0.582-0.835)

(0.555-0.760)

(0.560-0.731)

(0.200-0.275)

(0.185-0.338)

(0.474-0.718)

0.509

0.523

0.512

0.309

0.300

0.702

(0.356-0.679)

(0.371-0.656)

(0.406-0.636)

(0.239-0.379)

(0.209-0.409)

(0.425-0.893)

0.665

0.627

0.632

0.272

0.265

0.642

(0.521-0.798)

(0.516-0.731)

(0.552-0.729)

(0.210-0.338)

(0.171-0.377)

(0.357-0.909)

0.593

0.521

0.604

0.249

0.216

0.515

(0.465-0.696)

(0.445-0.592)

(0.518-0.670)

(0.221-0.283)

(0.139-0.307)

(0.312-0.679)

0.470

0.494

0.506

0.293

0.264

0.593

(0.356-0.601)

(0.412-0.596)

(0.426-0.596)

(0.246-0.358)

(0.171-0.354)

(0.371-0.780)

0.551

0.519

0.580

0.297

0.312

0.699

(0.424-0.684)

(0.421-0.599)

(0.476-0.669)

(0.233-0.374)

(0.236-0.432)

(0.458-0.927)

0.505

0.529

0.497

0.258

0.192

0.429

(0.413-0.591)

(0.480-0.583)

(0.431-0.566)

(0.240-0.274)

(0.133-0.262)

(0.264-0.602)

0.565

0.574

0.571

0.265

0.188

0.455

(0.466-0.635)

(0.507-0.650)

(0.508-0.620)

(0.232-0.297)

(0.120-0.274)

(0.289-0.623)

0.590

0.575

0.550

0.264

0.214

0.433

(0.491-0.674)

(0.511-0.658)

(0.481-0.616)

(0.236-0.292)

(0.146-0.289)

(0.291-0.567)

0.552

0.537

0.520

0.242

0.177

0.415

(0.424-0.678)

(0.463-0.610)

(0.428-0.602)

(0.211-0.274)

(0.110-0.252)

(0.249-0.644)

0.583

0.540

0.534

0.246

0.188

0.444

(0.456-0.731)

(0.433-0.646)

(0.454-0.611)

(0.197-0.291)

(0.119-0.271)

(0.261-0.641)

0.494

0.483

0.549

0.296

0.228

0.559

(0.356-0.641)

(0.431-0.545)

(0.469-0.626)

(0.242-0.353)

(0.150-0.313)

(0.354-0.832)

0.533

0.516

0.520

0.252

0.210

0.452

(0.430-0.645)

(0.420-0.620)

(0.430-0.591)

(0.229-0.274)

(0.132-0.285)

(0.285-0.611)

0.479

0.539

0.511

0.297

0.202

0.466

(0.337-0.613)

(0.482-0.607)

(0.428-0.594)

(0.240-0.363)

(0.113-0.290)

(0.253-0.703)

0.530

0.588

0.539

0.278

0.258

0.554

(0.396-0.676)

(0.498-0.671)

(0.461-0.623)

(0.233-0.324)

(0.159-0.348)

(0.335-0.746)

0.694

0.595

0.660

0.147

0.101

0.265

(0.607-0.789)

(0.507-0.687)

(0.580-0.751)

(0.120-0.183)

(0.061-0.152)

(0.140-0.434)

0.549

0.498

0.550

0.183

0.191

0.508

(0.437-0.649)

(0.398-0.600)

(0.453-0.640)

(0.140-0.237)

(0.140-0.268)

(0.266-0.804)

0.771

0.728

0.740

0.146

0.142

0.383

(0.696-0.846)

(0.635-0.810)

(0.676-0.805)

(0.116-0.194)

(0.095-0.202)

(0.240-0.610)

0.577

0.519

0.553

0.188

0.097

0.249

(0.520-0.637)

(0.453-0.583)

(0.496-0.604)

(0.166-0.213)

(0.061-0.136)

(0.136-0.446)

0.607

0.622

0.607

0.208

0.140

0.357

(0.548-0.665)

(0.554-0.680)

(0.546-0.663)

(0.183-0.234)

(0.087-0.181)

(0.235-0.498)

0.750

0.629

0.739

0.179

0.116

0.355

(0.694-0.805)

(0.577-0.692)

(0.688-0.785)

(0.145-0.220)

(0.076-0.162)

(0.194-0.588)

0.597

0.528

0.578

0.247

0.185

0.377

(0.516-0.677)

(0.444-0.606)

(0.499-0.643)

(0.239-0.256)

(0.126-0.254)

(0.265-0.532)

0.479

0.484

0.489

0.252

0.179

0.457

(0.365-0.577)

(0.392-0.561)

(0.399-0.570)

(0.225-0.283)

(0.125-0.238)

(0.299-0.609)

0.603

0.610

0.290

0.213

0.580

(0.503-0.679)

(0.525-0.680)

(0.541-0.669)

(0.249-0.329)

(0.150-0.262)

(0.404-0.814)

0.515

0.506

0.513

0.235

0.093

0.221

(0.459-0.578)

(0.460-0.560)

(0.469-0.562)

(0.222-0.245)

(0.060-0.133)

(0.142-0.336)

0.564

0.523

0.553

0.236

0.106

0.247

(0.511-0.625)

(0.483-0.560)

(0.502-0.603)

(0.219-0.252)

(0.067-0.141)

(0.171-0.401)

0.635

0.591

0.613

0.232

0.117

0.284

(0.578-0.683)

(0.538-0.634)

(0.572-0.653)

(0.211-0.257)

(0.077-0.158)

(0.192-0.426)

0.558

0.519

0.554

0.185

0.122

0.271

(0.462-0.648)

(0.456-0.587)

(0.472-0.637)

(0.158-0.208)

(0.080-0.169)

(0.172-0.432)

0.502

0.547

0.507

0.219

0.179

0.452

(0.395-0.585)

(0.487-0.611)

(0.427-0.575)

(0.183-0.254)

(0.130-0.232)

(0.280-0.637)

0.644

0.634

0.641

0.200

0.162

0.408

(0.554-0.737)

(0.566-0.709)

(0.570-0.710)

(0.167-0.235)

(0.112-0.216)

(0.247-0.583)

Continued on next page

Survival on Cisplatin for Bladder Cancer

Age-Sex Last Progress A POLLO

Survival on Cisplatin for Head & Neck Cancer

Age-Sex Last Progress A POLLO

Survival on Cytarabine for Acute Myeloid Leukemia

Age-Sex Last Progress A POLLO

Survival on Doxorubicin for Soft Tissue Sarcoma

Age-Sex Last Progress A POLLO

Survival on Enfortumab vedotin for Bladder Cancer

Age-Sex Last Progress A POLLO

Survival on Fluorouracil for Colorectal Cancer

Age-Sex Last Progress A POLLO

Survival on Gemcitabine for Pancreatic Cancer

Age-Sex Last Progress A POLLO

Survival on Ipilimumab for Kidney Cancer

Age-Sex Last Progress A POLLO

Survival on Leuprolide for Prostate Cancer

Age-Sex Last Progress A POLLO

Survival on Nivolumab for Kidney Cancer

Age-Sex Last Progress A POLLO

Survival on Melanoma

Nivolumab

for

Age-Sex Last Progress

0.455

0.441

0.465

0.195

0.166

0.355

(0.342-0.564)

(0.355-0.504)

(0.362-0.559)

(0.175-0.218)

(0.109-0.225)

(0.243-0.491)

0.399

0.389

0.411

0.220

0.216

0.521

(0.286-0.520)

(0.293-0.465)

(0.308-0.517)

(0.185-0.265)

(0.157-0.285)

(0.336-0.789)

0.658

0.645

0.669

0.179

0.138

0.332

(0.549-0.761)

(0.560-0.736)

(0.566-0.747)

(0.148-0.204)

(0.092-0.187)

(0.192-0.548)

0.554

0.574

0.548

0.102

0.064

0.139

(0.470-0.640)

(0.499-0.631)

(0.475-0.631)

(0.085-0.125)

(0.045-0.085)

(0.097-0.194)

0.582

0.530

0.583

0.108

0.071

0.176

(0.502-0.671)

(0.458-0.598)

(0.511-0.662)

(0.088-0.128)

(0.041-0.098)

(0.096-0.275)

0.751

0.712

0.748

0.099

0.060

0.219

(0.686-0.808)

(0.645-0.764)

(0.684-0.804)

(0.081-0.119)

(0.038-0.086)

(0.118-0.330)

0.508

0.477

0.513

0.243

0.142

0.328

(0.416-0.599)

(0.412-0.535)

(0.447-0.592)

(0.210-0.266)

(0.076-0.207)

(0.190-0.470)

0.584

0.560

0.588

0.244

0.131

0.316

(0.497-0.647)

(0.484-0.629)

(0.527-0.632)

(0.222-0.271)

(0.075-0.187)

(0.164-0.544)

0.651

0.598

0.647

0.223

0.153

0.425

(0.553-0.739)

(0.524-0.666)

(0.576-0.717)

(0.190-0.258)

(0.104-0.212)

(0.262-0.653)

0.657

0.550

0.621

0.258

0.222

0.467

(0.582-0.756)

(0.486-0.613)

(0.554-0.691)

(0.232-0.285)

(0.157-0.293)

(0.317-0.633)

0.634

0.606

0.621

0.256

0.218

0.462

(0.518-0.737)

(0.527-0.678)

(0.544-0.695)

(0.221-0.293)

(0.142-0.295)

(0.312-0.719)

0.714

0.642

0.673

0.240

0.206

0.472

(0.632-0.788)

(0.558-0.731)

(0.610-0.726)

(0.207-0.278)

(0.147-0.276)

(0.317-0.660)

nan

0.827

0.500

0.645

0.231

(0.660-0.953)

(0.500-0.500)

(0.560-0.769)

(0.173-0.283)

0.361

0.500

0.589

0.406

(0.093-0.653)

(0.500-0.500)

(0.468-0.707)

(0.242-0.549)

0.680

0.500

0.574

0.253

(0.441-0.915)

(0.500-0.500)

(0.461-0.695)

(0.135-0.368)

0.562

0.505

0.541

0.217

0.134

0.349

(0.478-0.624)

(0.472-0.533)

(0.478-0.589)

(0.193-0.245)

(0.093-0.172)

(0.234-0.503)

0.591

0.604

0.580

0.239

0.185

0.403

(0.512-0.668)

(0.533-0.666)

(0.520-0.638)

(0.204-0.271)

(0.145-0.223)

(0.288-0.570)

0.756

0.683

0.708

0.190

0.132

0.301

(0.694-0.818)

(0.632-0.740)

(0.652-0.760)

(0.159-0.223)

(0.085-0.178)

(0.206-0.434)

0.603

0.561

0.557

0.243

0.120

0.274

(0.524-0.683)

(0.492-0.633)

(0.504-0.620)

(0.231-0.252)

(0.070-0.182)

(0.166-0.426)

0.537

0.514

0.534

0.342

0.320

0.656

(0.442-0.625)

(0.442-0.599)

(0.469-0.590)

(0.305-0.387)

(0.253-0.397)

(0.493-0.874)

0.586

0.555

0.564

0.276

0.203

0.475

(0.497-0.659)

(0.474-0.620)

(0.504-0.611)

(0.245-0.312)

(0.150-0.255)

(0.320-0.657)

0.690

0.647

0.671

0.242

0.259

0.535

(0.556-0.806)

(0.537-0.767)

(0.565-0.762)

(0.202-0.272)

(0.164-0.348)

(0.333-0.687)

0.500

0.526

0.499

0.313

0.324

0.724

(0.363-0.632)

(0.462-0.595)

(0.410-0.609)

(0.239-0.395)

(0.207-0.425)

(0.477-0.962)

0.696

0.635

0.656

0.242

0.243

0.565

(0.549-0.836)

(0.514-0.768)

(0.573-0.742)

(0.180-0.311)

(0.162-0.352)

(0.344-0.812)

0.653

0.617

0.645

0.089

0.038

0.095

(0.590-0.719)

(0.551-0.668)

(0.583-0.705)

(0.074-0.109)

(0.024-0.055)

(0.052-0.160)

0.791

0.727

0.778

0.080

0.036

0.101

(0.733-0.841)

(0.681-0.775)

(0.729-0.826)

(0.061-0.098)

(0.016-0.056)

(0.042-0.175)

0.863

0.815

0.855

0.072

0.033

0.096

(0.816-0.909)

(0.762-0.872)

(0.811-0.893)

(0.056-0.088)

(0.018-0.051)

(0.044-0.172)

0.598

0.582

0.610

0.275

0.229

0.538

(0.487-0.701)

(0.503-0.664)

(0.528-0.682)

(0.233-0.316)

(0.153-0.300)

(0.353-0.711)

0.553

0.528

0.530

0.274

0.215

0.486

(0.428-0.654)

(0.423-0.623)

(0.457-0.611)

(0.217-0.318)

(0.128-0.290)

(0.275-0.719)

0.659

0.587

0.627

0.272

0.245

0.593

(0.548-0.764)

(0.517-0.669)

(0.551-0.721)

(0.212-0.329)

(0.166-0.329)

(0.342-0.802)

0.646

0.590

0.615

0.194

0.123

0.244

(0.573-0.716)

(0.522-0.659)

(0.555-0.673)

(0.178-0.207)

(0.075-0.161)

(0.169-0.324)

0.641

0.597

0.616

0.192

0.114

0.260

(0.534-0.715)

(0.520-0.666)

(0.529-0.686)

(0.170-0.216)

(0.077-0.174)

(0.159-0.411)

Continued on next page

A POLLO Survival on Octreotide for Neuroendocrine Tumors

Age-Sex Last Progress A POLLO

Survival on Osimertinib for Lung Cancer

Age-Sex Last Progress A POLLO

Survival on Oxaliplatin for Stomach Cancer

Age-Sex Last Progress A POLLO

Survival on Paclitaxel for Breast Cancer

Age-Sex Last Progress A POLLO

Survival on Paclitaxel for Ovarian Cancer

Age-Sex Last Progress A POLLO

Survival on Paclitaxel for Pancreatic Cancer

Age-Sex Last Progress A POLLO

Survival on Pembrolizumab for Breast Cancer

Age-Sex Last Progress A POLLO

Survival on Pembrolizumab for Colorectal Cancer

Age-Sex Last Progress A POLLO

Survival on Pembrolizumab for Lung Cancer

Age-Sex Last Progress A POLLO

Survival on Pembrolizumab for Melanoma

Age-Sex Last Progress A POLLO

Survival on Mesothelioma

Pemetrexed

for

Age-Sex

0.713

0.644

0.689

0.194

0.145

0.368

(0.637-0.773)

(0.566-0.707)

(0.630-0.740)

(0.170-0.217)

(0.101-0.183)

(0.224-0.520)

0.580

0.566

0.576

0.150

0.123

0.278

(0.396-0.696)

(0.426-0.654)

(0.415-0.686)

(0.120-0.186)

(0.069-0.188)

(0.159-0.520)

0.600

0.584

0.552

0.176

0.207

0.549

(0.470-0.724)

(0.482-0.676)

(0.446-0.660)

(0.140-0.222)

(0.145-0.282)

(0.351-0.704)

0.719

0.598

0.700

0.146

0.164

0.395

(0.588-0.839)

(0.469-0.709)

(0.577-0.806)

(0.109-0.184)

(0.111-0.235)

(0.244-0.701)

0.449

0.537

0.470

0.269

0.231

0.544

(0.325-0.570)

(0.431-0.657)

(0.374-0.582)

(0.247-0.289)

(0.149-0.330)

(0.412-0.631)

0.647

0.569

0.635

0.235

0.196

0.474

(0.522-0.790)

(0.480-0.693)

(0.539-0.736)

(0.183-0.276)

(0.119-0.286)

(0.257-0.735)

0.621

0.666

0.614

0.249

0.232

0.539

(0.487-0.725)

(0.572-0.759)

(0.517-0.689)

(0.205-0.301)

(0.149-0.309)

(0.306-0.804)

0.575

0.489

0.507

0.255

0.180

0.416

(0.488-0.675)

(0.450-0.531)

(0.448-0.581)

(0.244-0.268)

(0.110-0.250)

(0.222-0.540)

0.601

0.574

0.587

0.254

0.175

0.420

(0.506-0.703)

(0.492-0.656)

(0.527-0.659)

(0.227-0.286)

(0.117-0.242)

(0.269-0.582)

0.680

0.620

0.651

0.239

0.186

0.443

(0.591-0.773)

(0.548-0.692)

(0.593-0.707)

(0.204-0.277)

(0.135-0.259)

(0.279-0.598)

0.755

0.695

0.713

0.121

0.064

0.154

(0.693-0.816)

(0.645-0.750)

(0.654-0.775)

(0.104-0.141)

(0.037-0.096)

(0.083-0.288)

0.817

0.747

0.779

0.106

0.060

0.170

(0.749-0.863)

(0.694-0.802)

(0.724-0.817)

(0.084-0.130)

(0.037-0.095)

(0.090-0.276)

0.883

0.810

0.847

0.092

0.065

0.235

(0.840-0.921)

(0.756-0.861)

(0.813-0.881)

(0.074-0.112)

(0.038-0.088)

(0.100-0.367)

0.542

0.535

0.542

0.205

0.121

0.283

(0.455-0.618)

(0.454-0.597)

(0.464-0.606)

(0.182-0.232)

(0.066-0.169)

(0.173-0.446)

0.653

0.652

0.609

0.273

0.284

0.557

(0.582-0.751)

(0.597-0.724)

(0.553-0.694)

(0.225-0.331)

(0.235-0.339)

(0.393-0.736)

0.768

0.723

0.734

0.163

0.109

0.268

(0.667-0.851)

(0.650-0.801)

(0.652-0.805)

(0.124-0.211)

(0.068-0.154)

(0.158-0.414)

0.589

0.558

0.597

0.240

0.157

0.347

(0.503-0.703)

(0.471-0.638)

(0.540-0.665)

(0.218-0.263)

(0.098-0.221)

(0.215-0.486)

0.542

0.545

0.530

0.318

0.291

0.592

(0.437-0.639)

(0.469-0.632)

(0.460-0.593)

(0.273-0.368)

(0.234-0.352)

(0.428-0.776)

0.564

0.568

0.581

0.278

0.211

0.498

(0.453-0.657)

(0.486-0.649)

(0.525-0.639)

(0.245-0.317)

(0.144-0.280)

(0.331-0.689)

0.730

0.632

0.682

0.200

0.151

0.341

(0.630-0.813)

(0.549-0.715)

(0.619-0.752)

(0.173-0.234)

(0.100-0.221)

(0.236-0.516)

0.740

0.570

0.699

0.190

0.135

0.344

(0.627-0.835)

(0.485-0.645)

(0.626-0.774)

(0.156-0.232)

(0.084-0.186)

(0.192-0.620)

0.752

0.576

0.738

0.180

0.116

0.320

(0.639-0.840)

(0.494-0.647)

(0.651-0.809)

(0.132-0.228)

(0.056-0.179)

(0.157-0.540)

0.444

0.538

0.477

0.253

0.241

0.514

(0.282-0.626)

(0.455-0.612)

(0.374-0.590)

(0.236-0.269)

(0.164-0.332)

(0.353-0.590)

0.519

0.500

0.556

0.294

0.271

0.668

(0.372-0.690)

(0.500-0.500)

(0.467-0.657)

(0.212-0.375)

(0.160-0.399)

(0.384-0.934)

0.598

0.557

0.578

0.258

0.264

0.635

(0.455-0.747)

(0.444-0.687)

(0.463-0.696)

(0.191-0.316)

(0.174-0.357)

(0.393-0.824)

0.528

0.547

0.507

0.244

0.118

0.281

(0.440-0.614)

(0.474-0.615)

(0.440-0.572)

(0.233-0.255)

(0.067-0.178)

(0.153-0.428)

0.582

0.571

0.564

0.250

0.132

0.344

(0.500-0.658)

(0.506-0.635)

(0.501-0.619)

(0.223-0.276)

(0.086-0.186)

(0.213-0.521)

0.674

0.655

0.644

0.230

0.127

0.308

(0.609-0.759)

(0.589-0.725)

(0.585-0.705)

(0.200-0.256)

(0.081-0.193)

(0.170-0.523)

0.562

0.514

0.568

0.174

0.149

0.346

(0.442-0.656)

(0.424-0.598)

(0.458-0.652)

(0.156-0.195)

(0.095-0.195)

(0.230-0.425)

0.610

0.583

0.602

0.194

0.163

0.449

(0.513-0.709)

(0.507-0.656)

(0.519-0.691)

(0.169-0.215)

(0.112-0.207)

(0.281-0.618)

0.660

0.605

0.665

0.181

0.134

0.428

(0.577-0.743)

(0.536-0.677)

(0.597-0.729)

(0.147-0.216)

(0.081-0.189)

(0.228-0.614)

0.507

0.558

0.522

0.279

0.266

0.585

(0.349-0.646)

(0.514-0.617)

(0.399-0.619)

(0.241-0.326)

(0.188-0.369)

(0.400-0.721)

Continued on next page

Last Progress A POLLO Survival on Pertuzumab for Breast Cancer

Age-Sex Last Progress A POLLO

Survival on Rituximab for Diffuse Large B-Cell Lymphoma

Age-Sex Last Progress A POLLO

Survival on Rituximab for NonHodgkin Lymphoma

Age-Sex Last Progress A POLLO

Survival on Temozolomide for Glioblastoma

Age-Sex Last Progress A POLLO

Survival on Trastuzumab for Breast Cancer

Age-Sex Last Progress A POLLO

Exacerbation-Free Survival Tiotropium for COPD

Age-Sex Last Progress A POLLO

Exacerbation-Free Survival Umeclidinium / Vilanterol COPD

on for

Age-Sex

Last Progress A POLLO

0.571

0.474

0.591

0.291

0.257

0.637

(0.378-0.771)

(0.379-0.551)

(0.473-0.704)

(0.218-0.389)

(0.161-0.350)

(0.337-0.952)

0.550

0.494

0.591

0.274

0.218

0.529

(0.404-0.671)

(0.445-0.535)

(0.490-0.664)

(0.231-0.333)

(0.135-0.291)

(0.299-0.772)

0.746

0.646

0.737

0.051

0.065

0.136

(0.649-0.849)

(0.581-0.710)

(0.645-0.836)

(0.033-0.069)

(0.048-0.084)

(0.101-0.174)

0.837

0.748

0.835

0.056

0.061

0.211

(0.733-0.917)

(0.644-0.835)

(0.732-0.905)

(0.039-0.076)

(0.041-0.087)

(0.092-0.360)

0.892

0.853

0.891

0.049

0.037

0.166

(0.819-0.945)

(0.767-0.912)

(0.822-0.935)

(0.034-0.069)

(0.018-0.055)

(0.055-0.295)

0.586

0.550

0.576

0.146

0.103

0.236

(0.495-0.680)

(0.478-0.631)

(0.492-0.658)

(0.120-0.173)

(0.069-0.142)

(0.133-0.340)

0.517

0.519

0.524

0.189

0.164

0.470

(0.440-0.605)

(0.432-0.602)

(0.453-0.607)

(0.149-0.230)

(0.121-0.204)

(0.272-0.670)

0.712

0.691

0.696

0.159

0.127

0.473

(0.615-0.803)

(0.631-0.762)

(0.616-0.784)

(0.128-0.203)

(0.093-0.167)

(0.287-0.674)

0.686

0.619

0.643

0.166

0.075

0.189

(0.615-0.757)

(0.564-0.675)

(0.586-0.694)

(0.141-0.187)

(0.043-0.114)

(0.093-0.306)

0.559

0.544

0.563

0.198

0.142

0.392

(0.487-0.642)

(0.482-0.608)

(0.498-0.631)

(0.167-0.231)

(0.103-0.181)

(0.226-0.558)

0.732

0.684

0.700

0.153

0.096

0.258

(0.670-0.794)

(0.634-0.753)

(0.650-0.749)

(0.127-0.182)

(0.070-0.142)

(0.160-0.448)

0.583

0.587

0.564

0.244

0.178

0.409

(0.475-0.694)

(0.490-0.689)

(0.474-0.652)

(0.223-0.266)

(0.112-0.244)

(0.244-0.644)

0.612

0.556

0.574

0.247

0.202

0.486

(0.489-0.733)

(0.473-0.650)

(0.474-0.662)

(0.205-0.291)

(0.137-0.274)

(0.283-0.709)

0.651

0.615

0.619

0.237

0.214

0.497

(0.537-0.759)

(0.516-0.726)

(0.525-0.703)

(0.194-0.279)

(0.124-0.305)

(0.297-0.754)

0.657

0.553

0.639

0.100

0.048

0.102

(0.592-0.726)

(0.512-0.593)

(0.583-0.697)

(0.085-0.112)

(0.035-0.066)

(0.075-0.135)

0.899

0.843

0.866

0.064

0.058

0.129

(0.850-0.930)

(0.792-0.881)

(0.824-0.893)

(0.054-0.074)

(0.040-0.075)

(0.081-0.198)

0.928

0.866

0.896

0.053

0.054

0.167

(0.893-0.958)

(0.814-0.901)

(0.864-0.925)

(0.043-0.061)

(0.035-0.075)

(0.080-0.295)

0.588

0.538

0.577

0.231

0.088

0.196

(0.523-0.638)

(0.493-0.583)

(0.532-0.616)

(0.221-0.243)

(0.053-0.118)

(0.126-0.283)

0.650

0.596

0.602

0.225

0.094

0.213

(0.597-0.708)

(0.551-0.637)

(0.554-0.641)

(0.205-0.244)

(0.059-0.128)

(0.123-0.317)

0.750

0.684

0.699

0.198

0.091

0.215

(0.701-0.800)

(0.640-0.729)

(0.663-0.734)

(0.181-0.218)

(0.059-0.126)

(0.132-0.350)

0.598

0.522

0.542

0.230

0.140

0.336

(0.492-0.708)

(0.456-0.594)

(0.472-0.625)

(0.209-0.254)

(0.080-0.203)

(0.178-0.549)

0.630

0.586

0.607

0.233

0.156

0.368

(0.528-0.728)

(0.505-0.674)

(0.546-0.679)

(0.204-0.268)

(0.090-0.212)

(0.230-0.581)

0.706

0.618

0.689

0.210

0.142

0.359

(0.595-0.806)

(0.549-0.694)

(0.620-0.753)

(0.169-0.256)

(0.081-0.218)

(0.172-0.558)

Extended Data Table 12: Definition of adverse event prediction tasks. We evaluated TTE performance on 17 adverse event tasks. τ is the threshold duration. Incidence refers to the percentage of patients in the test set who reach the endpoint by the specified τ . The Snapshot Event column specifies all medications used to identify the treatment indicated in the Task column. Integers refer to ingredient RxCUIs, and the ∧ symbol indicates combination therapies where both ingredients must be present. The Endpoint Event column specifies all ICD-10 codes used to identify the disease indicated in the Task column. Category

Task

Incid.

Snapshot Event

Endpoint Event

Cardiovascular & Circulatory

Angioedema on ACE Inhibitors

90d

37,960

0.05%

T78.3

Cardiovascular & Circulatory

Cardiomyopathy After Anthracyclines

6,254

9.28%

Cardiovascular & Circulatory

MACE on Levothyroxine with History of Heart Failure Major Bleeding on Oral Anticoagulants

1,805

36.45%

Lisinopril, Zestril, Prinivil, Enalapril, Vasotec, Ramipril, Altace, Benazepril, Lotensin, Captopril, Capoten, Fosinopril, Monopril, Quinapril, Accupril, Perindopril, Aceon, Trandolapril, Mavik, Moexipril, Univasc doxorubicin, Adriamycin, Doxil, Caelyx, Myocet, Lipodox, Rubex, Celdoxome Pegylated Liposomal, Zolsketil Pegylated Liposomal, epirubicin, Ellence, Pharmorubicin, Farmorubicin, daunorubicin, Cerubidine, DaunoXome, Vyxeos, idarubicin, Idamycin, Zavedos, valrubicin, Valstar, aclarubicin, Aclacinon, Aclacin, pirarubicin, Pinorubin, Therarubicin, amrubicin, Calsed 10582

18,723

2.72%

Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive

Hyperthyroidism on Amiodarone Hypoglycemia on Sulfonylureas

9,870

0.79%

4,620

1.04%

Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive

Hypothyroidism on Amiodarone Type 2 Diabetes After Atypical Antipsychotics

4,805

6.26%

16,390

3.03%

Cardiovascular & Circulatory

Warfarin, Coumadin, Jantoven, Apixaban, Eliquis, Rivaroxaban, Xarelto, Edoxaban, Savaysa, Dabigatran etexilate, Pradaxa Amiodarone, Cordarone, Pacerone, Nexterone Glipizide, Glucotrol, Glyburide, Diabeta, Micronase, Glibenclamide, Glimepiride, Amaryl, Chlorpropamide, Tolazamide, Tolbutamide Amiodarone, Cordarone, Pacerone, Nexterone Olanzapine, Zyprexa, Risperidone, Risperdal, Paliperidone, Invega, Quetiapine, Seroquel, Ziprasidone, Geodon, Aripiprazole, Abilify, Aristada, Brexpiprazole, Rexulti, Cariprazine, Vraylar, Lurasidone, Latuda, Iloperidone, Fanapt, Asenapine, Saphris, Secuado, Lumateperone, Caplyta

I42, I50

I21, I22, I60, I61, I62, I63, death D62, I60, I61, I62, I85.01, I97.4, I97.6, K25.0, K26.0, K29.01, K92.2, N02, R04.2, R04.89, R58 E05 E16.0, E16.A

E16.1,

E16.2,

E02, E03 E11

Continued on next page

Endocrine, Metabolic & Reproductive

Type 2 Diabetes After Systemic Corticosteroids

135,146

1.87%

Gastrointestinal, Hepatic & Renal

Acute Kidney Injury After NSAIDs

30d

137,188

0.25%

Gastrointestinal, Hepatic & Renal

GI Bleeding on Antiplatelet Agents

55,326

0.85%

Gastrointestinal, Hepatic & Renal

GI Bleeding on NSAIDs

90d

133,808

0.10%

Hematology & Sensory

Heparin-Induced Thrombocytopenia

14d

141,647

0.01%

Hematology & Sensory

Hyperkalemia on ACE Inhibitors

30d

20,483

0.41%

Prednisone, Deltasone, Rayos, Sterapred, Prednisolone, Orapred, Prelone, Pediapred, Millipred, Methylprednisolone, Medrol, Hydrocortisone, Cortef, Hydrocortone, SoluCortef, Dexamethasone, Decadron, DexPak, Ozurdex, Maxidex, Betamethasone, Celestone, Celestone Soluspan, Diprolene, Triamcinolone, Kenalog, Aristocort, Nasacort, Hydrocort Ibuprofen, Advil, Motrin, Naproxen, Aleve, Naprosyn, Diclofenac, Voltaren, Cataflam, Indomethacin, Indocin, Meloxicam, Mobic, Celecoxib, Celebrex, Etodolac, Lodine, Ketorolac, Toradol, Nabumetone, Relafen, Piroxicam, Feldene, Sulindac, Oxaprozin, Daypro, Mefenamic acid, Ponstel, Flurbiprofen, Ansaid Aspirin, Acetylsalicylic acid, ASA, Clopidogrel, Plavix, Prasugrel, Effient, Ticagrelor, Brilinta, Cangrelor, Kengreal, Dipyridamole, Persantine, Aggrenox, Ticlopidine, Ticlid Ibuprofen, Advil, Motrin, Naproxen, Aleve, Naprosyn, Diclofenac, Voltaren, Cataflam, Indomethacin, Indocin, Meloxicam, Mobic, Celecoxib, Celebrex, Etodolac, Lodine, Ketorolac, Toradol, Nabumetone, Relafen, Piroxicam, Feldene, Sulindac, Oxaprozin, Daypro, Mefenamic acid, Ponstel, Flurbiprofen, Ansaid Heparin, Unfractionated heparin, UFH, Enoxaparin, Lovenox, Dalteparin, Fragmin, Tinzaparin, Innohep, Fondaparinux, Arixtra Lisinopril, Zestril, Prinivil, Enalapril, Vasotec, Ramipril, Altace, Benazepril, Lotensin, Captopril, Capoten, Fosinopril, Monopril, Quinapril, Accupril, Perindopril, Aceon, Trandolapril, Mavik, Moexipril, Univasc

E11

N17

K92.2

D75.82

E87.5

Continued on next page

Musculoskeletal, Autoimmune & Dermatologic

Rhabdomyolysis on Statins

75,872

0.08%

Neurology & Psychiatry

Tardive Dyskinesia on Antipsychotics

81,759

0.07%

Respiratory & Infectious

Clostridioides difficile Infection on PPIs

45,097

0.49%

Atorvastatin, Lipitor, Rosuvastatin, Crestor, Simvastatin, Zocor, Pravastatin, Pravachol, Lovastatin, Mevacor, Altoprev, Fluvastatin, Lescol, Pitavastatin, Livalo Haloperidol, Haldol, Fluphenazine, Prolixin, Modecate, Trifluoperazine, Stelazine, Thiothixene, Navane, Pimozide, Orap, Loxapine, Loxitane, Perphenazine, Trilafon, Chlorpromazine, Thorazine, Largactil, Thioridazine, Mellaril, Mesoridazine, Serentil, Olanzapine, Zyprexa, Risperidone, Risperdal, Paliperidone, Invega, Quetiapine, Seroquel, Ziprasidone, Geodon, Aripiprazole, Abilify, Aristada, Brexpiprazole, Rexulti, Cariprazine, Vraylar, Lurasidone, Latuda, Iloperidone, Fanapt, Asenapine, Saphris, Secuado, Lumateperone, Caplyta Omeprazole, Prilosec, Esomeprazole, Nexium, Lansoprazole, Prevacid, Dexlansoprazole, Dexilant, Pantoprazole, Protonix, Rabeprazole, AcipHex

M62.82

G24.0

A04.7

Extended Data Table 13: Evaluation results of adverse event prediction tasks. We report time-dependent performance metrics computed at τ on the test set for 17 adverse event tasks. τ is the threshold duration. Arrows indicate whether higher (↑) or lower (↓) values are better. ICI, Integrated Calibration Index; MCE, Maximum Calibration Error. 10−4 is the default value of the penalizer (λ) for Cox proportional hazards; if convergence cannot be achieved then λ is increased to 1e-3 (∗ ) or 1e-2 (∗∗ ). Mean and 95% CI reported over 100 bootstraps of the test set. Task

Model

Angioedema on ACE Inhibitors

90d

Age-Sex Last Progress A POLLO

Cardiomyopathy After Anthracyclines

Age-Sex Last Progress A POLLO

MACE on Levothyroxine with History of Heart Failure

Age-Sex Last Progress A POLLO

Major Bleeding on Oral Anticoagulants

Age-Sex Last Progress A POLLO

Hyperthyroidism on Amiodarone

Age-Sex Last Progress A POLLO

Hypoglycemia on Sulfonylureas

Age-Sex Last Progress A POLLO

Hypothyroidism on Amiodarone

Age-Sex Last Progress A POLLO

Type 2 Diabetes After Atypical Antipsychotics

Age-Sex Last Progress A POLLO

Type 2 Diabetes After Systemic Corticosteroids

Age-Sex Last Progress A POLLO

Acute Kidney Injury After NSAIDs

30d

Age-Sex

AUC (↑)

Bal. Acc. (↑)

C-Index (↑)

Brier Score (↓)

ICI (↓)

MCE (↓)

0.600

0.592

0.603

0.001

0.000

0.001

(0.524-0.719)

(0.499-0.672)

(0.527-0.720)

(0.000-0.001)

(0.001-0.003)

0.556

0.500

0.560

0.000

0.001

(0.420-0.683)

(0.500-0.501)

(0.426-0.686)

(0.000-0.001)

(0.001-0.002)

nan

0.821

0.500

0.824

0.000

(0.680-0.931)

(0.500-0.500)

(0.685-0.932)

(0.000-0.000)

0.714

0.682

0.683

0.041

0.021

0.049

(0.662-0.771)

(0.630-0.733)

(0.633-0.738)

(0.035-0.049)

(0.014-0.028)

(0.032-0.081)

0.592

0.599

0.566

0.044

0.034

0.099

(0.509-0.659)

(0.551-0.649)

(0.483-0.636)

(0.036-0.053)

(0.025-0.043)

(0.056-0.132)

0.771

0.723

0.741

0.040

0.021

0.066

(0.716-0.817)

(0.677-0.767)

(0.687-0.780)

(0.032-0.050)

(0.013-0.032)

(0.031-0.126)

0.670

0.625

0.633

0.202

0.077

0.173

(0.622-0.722)

(0.577-0.671)

(0.594-0.680)

(0.190-0.215)

(0.041-0.105)

(0.108-0.247)

0.558

0.556

0.552

0.229

0.106

0.269

(0.504-0.618)

(0.505-0.597)

(0.511-0.594)

(0.210-0.246)

(0.073-0.139)

(0.166-0.394)

0.712

0.658

0.680

0.190

0.063

0.159

(0.662-0.762)

(0.608-0.705)

(0.641-0.721)

(0.171-0.205)

(0.035-0.095)

(0.085-0.240)

0.610

0.569

0.604

0.023

0.006

0.013

(0.564-0.658)

(0.531-0.613)

(0.558-0.650)

(0.020-0.026)

(0.004-0.008)

(0.008-0.023)

0.605

0.563

0.601

0.023

0.009

0.027

(0.561-0.643)

(0.533-0.592)

(0.557-0.638)

(0.019-0.027)

(0.005-0.012)

(0.014-0.041)

0.675

0.624

0.673

0.023

0.008

0.031

(0.644-0.711)

(0.594-0.662)

(0.643-0.709)

(0.020-0.026)

(0.005-0.011)

(0.017-0.044)

0.528

0.540

0.522

0.006

0.005

0.009

(0.414-0.635)

(0.444-0.622)

(0.410-0.631)

(0.004-0.009)

(0.003-0.006)

(0.007-0.013)

0.582

0.596

0.588

0.006

0.007

0.033

(0.450-0.712)

(0.498-0.702)

(0.457-0.720)

(0.004-0.008)

(0.006-0.009)

(0.021-0.042)

0.673

0.649

0.688

0.007

0.034

(0.568-0.777)

(0.551-0.748)

(0.587-0.786)

(0.005-0.009)

(0.022-0.044)

0.514

0.512

0.506

0.009

0.007

0.021

(0.422-0.603)

(0.407-0.597)

(0.415-0.594)

(0.006-0.013)

(0.004-0.011)

(0.010-0.036)

0.649

0.600

0.645

0.009

0.007

0.020

(0.567-0.736)

(0.515-0.674)

(0.562-0.732)

(0.005-0.014)

(0.003-0.011)

(0.008-0.036)

0.693

0.542

0.693

0.009

0.006

0.018

(0.590-0.811)

(0.477-0.615)

(0.589-0.811)

(0.005-0.014)

(0.003-0.010)

(0.009-0.030)

0.635

0.601

0.624

0.043

0.024

0.064

(0.578-0.698)

(0.536-0.656)

(0.567-0.682)

(0.034-0.052)

(0.016-0.032)

(0.032-0.104)

0.626

0.621

0.630

0.048

0.031

0.114

(0.522-0.708)

(0.551-0.674)

(0.527-0.706)

(0.037-0.058)

(0.022-0.044)

(0.060-0.182)

0.638

0.611

0.645

0.048

0.032

0.106

(0.548-0.707)

(0.546-0.662)

(0.560-0.707)

(0.038-0.056)

(0.023-0.042)

(0.052-0.159)

0.693

0.653

0.673

0.021

0.011

0.027

(0.646-0.729)

(0.605-0.700)

(0.625-0.710)

(0.018-0.024)

(0.008-0.015)

(0.015-0.043)

0.620

0.563

0.612

0.022

0.010

0.027

(0.557-0.676)

(0.519-0.603)

(0.550-0.669)

(0.018-0.026)

(0.007-0.014)

(0.016-0.039)

0.841

0.752

0.836

0.020

0.006

0.019

(0.801-0.884)

(0.702-0.798)

(0.795-0.878)

(0.017-0.024)

(0.003-0.009)

(0.008-0.038)

0.697

0.639

0.693

0.014

0.003

0.006

(0.679-0.714)

(0.627-0.648)

(0.675-0.710)

(0.013-0.015)

(0.002-0.003)

(0.003-0.009)

0.653

0.600

0.651

0.014

0.002

0.007

(0.626-0.676)

(0.585-0.615)

(0.624-0.674)

(0.013-0.015)

(0.001-0.003)

(0.003-0.012)

0.866

0.783

0.863

0.014

0.002

0.009

(0.850-0.878)

(0.768-0.798)

(0.848-0.875)

(0.013-0.015)

(0.002-0.004)

(0.004-0.017)

0.796

0.721

0.796

0.002

0.001

0.003

(0.761-0.827)

(0.694-0.748)

(0.761-0.827)

(0.002-0.003)

(0.001-0.001)

(0.002-0.005)

Continued on next page

Last Progress A POLLO GI Bleeding on Antiplatelet Agents

Age-Sex Last Progress A POLLO

GI Bleeding on NSAIDs

90d

Age-Sex Last Progress A POLLO

Heparin-Induced Thrombocytopenia

14d

Age-Sex Last Progress A POLLO

Hyperkalemia on ACE Inhibitors

30d

Age-Sex Last Progress A POLLO

Rhabdomyolysis on Statins

Age-Sex Last Progress A POLLO

Tardive Dyskinesia on Antipsychotics

Age-Sex Last Progress A POLLO

Clostridioides difficile Infection on PPIs

Age-Sex Last Progress A POLLO

0.838

0.731

0.837

0.002

0.001

0.002

(0.806-0.866)

(0.683-0.768)

(0.805-0.864)

(0.002-0.003)

(0.000-0.001)

(0.001-0.005)

0.906

0.830

0.905

0.002

0.001

0.003

(0.889-0.921)

(0.805-0.855)

(0.889-0.920)

(0.002-0.003)

(0.000-0.001)

(0.001-0.006)

0.693

0.629

0.689

0.007

0.002

0.005

(0.658-0.730)

(0.586-0.669)

(0.654-0.725)

(0.007-0.009)

(0.001-0.003)

(0.003-0.008)

0.689

0.652

0.685

0.008

0.003

0.012

(0.654-0.724)

(0.623-0.685)

(0.650-0.719)

(0.007-0.009)

(0.003-0.004)

(0.007-0.017)

0.802

0.726

0.799

0.007

0.002

0.010

(0.772-0.826)

(0.686-0.750)

(0.769-0.823)

(0.006-0.008)

(0.001-0.004)

(0.004-0.017)

0.804

0.716

0.804

0.001

0.000

0.001

(0.760-0.845)

(0.665-0.746)

(0.760-0.844)

(0.001-0.001)

(0.000-0.001)

(0.000-0.002)

0.793

0.723

0.791

0.001

0.000

0.002

(0.728-0.848)

(0.664-0.785)

(0.727-0.846)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.905

0.797

0.905

0.000

0.001

(0.842-0.951)

(0.681-0.904)

(0.842-0.950)

(0.000-0.000)

(0.000-0.002)

0.449

0.603

0.447

0.000

(0.299-0.534)

(0.418-0.736)

(0.298-0.533)

(0.000-0.000)

(0.000-0.001)

0.811

0.700

0.810

0.000

(0.690-0.941)

(0.437-0.850)

(0.688-0.941)

(0.000-0.000)

0.743

0.599

0.741

0.000

0.001

(0.542-0.949)

(0.482-0.799)

(0.538-0.948)

(0.000-0.000)

(0.000-0.001)

0.581

0.542

0.580

0.004

0.002

0.006

(0.476-0.668)

(0.466-0.613)

(0.476-0.667)

(0.003-0.005)

(0.001-0.003)

(0.003-0.010)

0.608

0.547

0.606

0.004

0.003

0.008

(0.516-0.698)

(0.458-0.622)

(0.514-0.696)

(0.003-0.005)

(0.001-0.004)

(0.004-0.014)

0.734

0.651

0.734

0.004

0.002

0.007

(0.662-0.801)

(0.569-0.722)

(0.663-0.801)

(0.003-0.005)

(0.001-0.003)

(0.003-0.012)

0.608

0.547

0.604

0.001

(0.502-0.703)

(0.481-0.629)

(0.498-0.698)

(0.000-0.001)

(0.001-0.001)

0.621

0.598

0.620

0.001

0.000

0.001

(0.521-0.731)

(0.516-0.693)

(0.520-0.729)

(0.000-0.001)

(0.001-0.002)

0.709

0.601

0.706

0.001

0.002

(0.636-0.781)

(0.528-0.686)

(0.632-0.777)

(0.000-0.001)

(0.001-0.003)

0.549

0.525

0.544

0.001

0.000

0.001

(0.446-0.653)

(0.417-0.612)

(0.443-0.647)

(0.000-0.001)

(0.001-0.002)

0.842

0.733

0.842

0.001

0.000

0.001

(0.781-0.890)

(0.639-0.822)

(0.779-0.890)

(0.000-0.001)

(0.000-0.000)

(0.000-0.003)

0.832

0.756

0.837

0.001

0.000

0.001

(0.727-0.909)

(0.671-0.823)

(0.736-0.913)

(0.000-0.001)

(0.000-0.000)

(0.000-0.002)

0.537

0.521

0.532

0.004

0.003

0.005

(0.476-0.602)

(0.464-0.569)

(0.473-0.597)

(0.003-0.005)

(0.002-0.003)

(0.005-0.007)

0.662

0.633

0.653

0.004

0.003

0.013

(0.580-0.741)

(0.575-0.698)

(0.572-0.734)

(0.003-0.005)

(0.002-0.004)

(0.007-0.018)

0.740

0.685

0.738

0.004

0.002

0.013

(0.688-0.797)

(0.636-0.735)

(0.686-0.794)

(0.003-0.005)

(0.002-0.003)

(0.008-0.018)

Extended Data Table 14: Definition of hospital operations prediction tasks. We evaluated TTE performance on 12 hospital operations tasks. The patient history is given to the model up to the snapshot event. The model predicts whether the endpoint event (hospital admission, discharge, or disease diagnosis) will occur within the duration indicated by τ . Incidence refers to the percentage of patients in the test set who reach the endpoint by the specified τ . For tasks where the trigger event is a hospital admission, we set the snapshot time to 24 hours after admission to accumulate predictive features. For example, in the Acute Sepsis task, the model is given all data up to 24 hours after ED arrival and predicts whether the patient will be diagnosed with sepsis within the next 6 days (1 week after admission). Task

Snapshot Event

Endpoint Event

Acute Cardiac Arrest Acute Catheter-Associated UTI Acute Clostridioides difficile Infection Acute Dialysis Dependence Acute Kidney Failure Acute Liver Failure Acute Pressure Ulcer Acute Respiratory Distress Syndrome Acute Sepsis Acute Venous Thromboembolism Inpatient Length of Stay Inpatient Readmission (All Cause)

24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after ED Admission 24hr after Inpatient Admission Inpatient Discharge

6d 6d 6d 6d 6d 6d 6d 6d 6d 6d 6d 30d

I46 T83.511 A04.7 Z99.2 N17 K72.00, K72.01, K72.90, K72.91 L89 J80 A40, A41, O85, R65.20, R65.21, T81.12 I26, I81, I82 Inpatient Discharge Inpatient Admission

Incid.

140,994 89,598 135,902 107,664 114,095 129,304 105,170 276,203 124,037 130,197 138,132 41,289

0.01% 0.01% 0.05% 0.02% 0.51% 0.01% 0.07% 0.01% 0.10% 0.23% 81.95% 27.31%

Extended Data Table 15: Evaluation results of hospital operations prediction tasks. We report timedependent performance metrics computed at τ on the test set for 12 hospital operations tasks. τ is the threshold duration. Arrows indicate whether higher (↑) or lower (↓) values are better. ICI, Integrated Calibration Index; MCE, Maximum Calibration Error. 10−4 is the default value of the penalizer (λ) for Cox proportional hazards; if convergence cannot be achieved then λ is increased to 1e-3 (∗ ) or 1e-2 (∗∗ ). Mean and 95% CI reported over 100 bootstraps of the test set. Task

Model

Acute Cardiac Arrest

Age-Sex Last Progress A POLLO

Acute Catheter-Associated UTI

Age-Sex Last Progress A POLLO

Acute Clostridioides difficile Infection

Age-Sex Last Progress A POLLO

Acute Dialysis Dependence

Age-Sex Last Progress A POLLO

Acute Kidney Failure

Age-Sex Last Progress A POLLO

Acute Liver Failure

Age-Sex Last Progress∗ A POLLO

Acute Pressure Ulcer

Age-Sex Last Progress A POLLO

Acute Respiratory Distress Syndrome

Age-Sex Last Progress A POLLO

Acute Sepsis

Age-Sex Last Progress A POLLO

Acute Venous Thromboembolism

Age-Sex Last Progress

AUC (↑)

Bal. Acc. (↑)

C-Index (↑)

Brier Score (↓)

ICI (↓)

MCE (↓)

0.689

0.577

0.688

0.000

(0.566-0.848)

(0.404-0.779)

(0.565-0.848)

(0.000-0.000)

(0.000-0.001)

0.791

0.692

0.790

0.000

0.001

(0.444-0.969)

(0.499-0.775)

(0.442-0.968)

(0.000-0.000)

(0.000-0.001)

0.832

0.833

0.831

0.000

0.001

(0.633-0.969)

(0.671-0.922)

(0.632-0.968)

(0.000-0.000)

(0.000-0.001)

0.894

0.815

0.892

0.000

(0.808-0.956)

(0.589-0.925)

(0.806-0.955)

(0.000-0.000)

(0.000-0.001)

0.970

0.784

0.969

0.000

0.001

(0.945-0.989)

(0.781-0.787)

(0.943-0.989)

(0.000-0.000)

(0.000-0.002)

0.953

0.906

0.952

0.000

0.001

(0.920-0.991)

(0.904-0.908)

(0.919-0.991)

(0.000-0.000)

(0.000-0.002)

0.762

0.680

0.761

0.001

0.000

0.001

(0.698-0.836)

(0.602-0.761)

(0.697-0.835)

(0.000-0.001)

(0.001-0.003)

0.751

0.737

0.749

0.001

0.000

0.002

(0.645-0.857)

(0.672-0.790)

(0.644-0.855)

(0.000-0.001)

(0.001-0.004)

0.884

0.798

0.883

0.001

0.000

0.002

(0.838-0.929)

(0.736-0.858)

(0.837-0.929)

(0.000-0.001)

(0.000-0.000)

(0.001-0.003)

0.727

0.655

0.726

0.000

0.001

(0.613-0.833)

(0.519-0.738)

(0.612-0.832)

(0.000-0.000)

(0.000-0.001)

0.910

0.882

0.909

0.000

0.001

(0.803-0.969)

(0.789-0.921)

(0.801-0.969)

(0.000-0.001)

(0.000-0.000)

(0.000-0.002)

0.991

0.974

0.991

0.000

0.001

(0.985-0.996)

(0.973-0.975)

(0.985-0.996)

(0.000-0.000)

(0.000-0.002)

0.694

0.639

0.692

0.005

0.001

0.003

(0.668-0.721)

(0.611-0.667)

(0.666-0.719)

(0.004-0.005)

(0.001-0.001)

(0.001-0.004)

0.800

0.727

0.796

0.006

0.002

0.008

(0.772-0.821)

(0.701-0.747)

(0.767-0.816)

(0.005-0.007)

(0.001-0.002)

(0.003-0.012)

0.875

0.801

0.872

0.005

0.001

0.006

(0.855-0.889)

(0.786-0.816)

(0.852-0.887)

(0.004-0.006)

(0.001-0.002)

(0.003-0.011)

0.588

0.557

0.587

0.000

(0.321-0.826)

(0.258-0.757)

(0.320-0.825)

(0.000-0.000)

(0.000-0.001)

0.794

0.737

0.792

0.000

0.001

(0.444-0.997)

(0.393-0.895)

(0.441-0.997)

(0.000-0.000)

(0.000-0.001)

0.727

0.722

0.727

0.000

(0.359-0.996)

(0.459-0.932)

(0.358-0.996)

(0.000-0.000)

(0.000-0.001)

0.816

0.673

0.815

0.001

0.000

0.002

(0.749-0.875)

(0.605-0.737)

(0.748-0.874)

(0.000-0.001)

(0.001-0.003)

0.820

0.704

0.818

0.001

0.000

0.001

(0.767-0.885)

(0.624-0.783)

(0.765-0.883)

(0.001-0.001)

(0.000-0.001)

(0.001-0.003)

0.925

0.883

0.925

0.001

0.000

0.001

(0.890-0.948)

(0.852-0.898)

(0.890-0.948)

(0.000-0.001)

(0.000-0.000)

(0.000-0.002)

0.592

0.574

0.591

0.000

(0.459-0.737)

(0.439-0.722)

(0.458-0.736)

(0.000-0.000)

0.783

0.730

0.781

0.000

(0.571-0.937)

(0.592-0.882)

(0.569-0.936)

(0.000-0.000)

(0.000-0.001)

0.974

0.544

0.973

0.000

(0.957-0.985)

(0.498-0.633)

(0.956-0.984)

(0.000-0.000)

0.703

0.661

0.703

0.001

0.000

0.001

(0.626-0.771)

(0.597-0.726)

(0.625-0.770)

(0.001-0.001)

(0.000-0.001)

(0.001-0.002)

0.712

0.655

0.709

0.001

0.002

(0.639-0.789)

(0.594-0.720)

(0.636-0.786)

(0.001-0.002)

(0.000-0.001)

(0.001-0.003)

0.865

0.790

0.863

0.001

0.000

0.001

(0.833-0.895)

(0.736-0.840)

(0.831-0.893)

(0.001-0.001)

(0.000-0.000)

(0.000-0.003)

0.644

0.593

0.643

0.002

0.001

(0.605-0.687)

(0.566-0.624)

(0.604-0.686)

(0.002-0.003)

(0.000-0.001)

(0.001-0.003)

0.696

0.641

0.694

0.003

0.001

0.002

(0.652-0.741)

(0.600-0.674)

(0.650-0.739)

(0.002-0.003)

(0.000-0.001)

(0.001-0.004)

Continued on next page

A POLLO Inpatient Length of Stay

Age-Sex Last Progress A POLLO

Inpatient Readmission (All Cause)

30d

Age-Sex Last Progress A POLLO

0.832

0.744

0.831

0.002

0.001

0.002

(0.805-0.857)

(0.710-0.774)

(0.804-0.856)

(0.002-0.003)

(0.000-0.001)

(0.001-0.004)

0.630

0.627

0.571

0.143

0.033

0.074

(0.625-0.636)

(0.623-0.632)

(0.568-0.575)

(0.141-0.145)

(0.029-0.036)

(0.067-0.081)

0.820

0.741

0.669

0.121

0.022

0.067

(0.815-0.825)

(0.735-0.747)

(0.667-0.672)

(0.119-0.123)

(0.019-0.026)

(0.054-0.081)

0.841

0.762

0.678

0.113

0.019

0.074

(0.837-0.845)

(0.759-0.767)

(0.675-0.681)

(0.112-0.115)

(0.016-0.021)

(0.061-0.089)

0.541

0.536

0.534

0.198

0.026

0.092

(0.531-0.551)

(0.528-0.544)

(0.526-0.544)

(0.194-0.201)

(0.021-0.032)

(0.074-0.109)

0.664

0.611

0.647

0.186

0.031

0.076

(0.653-0.676)

(0.602-0.620)

(0.638-0.657)

(0.182-0.189)

(0.024-0.039)

(0.057-0.098)

0.670

0.624

0.647

0.184

0.017

0.035

(0.661-0.680)

(0.616-0.632)

(0.639-0.655)

(0.181-0.187)

(0.011-0.022)

(0.023-0.052)

Extended Data Table 16: Definition of retrieval cohorts. We evaluated retrieval performance on 61 cohorts. Each cohort is defined by patients diagnosed with the indicated disease (ICD-10 codes) and later prescribed the indicated therapy (ingredient RxCUIs or drug names). Cohort Size indicates the number of patients in each cohort. P is the cohort size divided by the total size of the evaluation set (1.4M patients). Category

Task

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory

AFib on Apixaban Acute MI on Aspirin Chronic Ischemic Heart Disease on Aspirin Heart Failure on Carvedilol Hypertension on Lisinopril Peripheral Artery Disease on Cilostazol Stroke on Alteplase VTE on Apixaban Gestational Diabetes on Insulin

Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Cardiovascular & Circulatory Endocrine, Metabolic & Reproductive

Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Endocrine, Metabolic & Reproductive Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Gastrointestinal, Hepatic & Renal Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Hematology & Sensory Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Musculoskeletal, Autoimmune & Dermatologic Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry

Cohort Size

Disease

Therapy

7,039 2,840 11,036

0.49% 0.2% 0.77%

I48 I21 I25

1364430 1191 1191

1,594 16,562 156

0.11% 1.2% 0.011%

20352 29046 21107

652 2,995 472

0.046% 0.21% 0.033%

I50 I10 I70.2, I70.3, I70.4, I70.5, I70.6, I70.7, I73.9 I60, I61, I62, I63 I26, I81, I82 O24.4

330

0.023%

E05

8410 1364430 insulin, Humalog, Admelog, Lyumjev, Novolog, Fiasp, Merilog, Trurapi, NovoRapid, Apidra, Humulin, Novolin, Myxredlin, Lantus, Basaglar, Toujeo, Semglee, Rezvoglar, Levemir, Tresiba 6835

Hyperthyroidism on Methimazole Hypothyroidism on Levothyroxine Lipidemia on Atorvastatin

11,973

0.84%

E02, E03

10582

21,012

1.5%

E78

83367

PCOS on Metformin

161

0.011%

E28.2

metformin, glucophage

Preeclampsia on Magnesium Sulfate T2DM on Metformin

390

0.027%

O14

6585

5,973

0.42%

E11

metformin, glucophage

BPH on Tamsulosin CKD on Lisinopril Cirrhosis on Propranolol Crohn’s Disease on Budesonide IBD on Mesalamine Ulcerative Colitis on Mesalamine B12 Deficiency Anemia on Cyanocobalamin Glaucoma on Latanoprost Immune Thrombocytopenia on Prednisone Iron Deficiency Anemia on Ferrous Sulfate Macular Degeneration on Ranibizumab Gout on Allopurinol

3,879 2,768 141 125

0.27% 0.19% 0.0099% 0.0087%

N40 N18 K70.3, K71.7, K74 K50

77492 29046 8787 19831

426 276

0.03% 0.019%

K50, K51, K52.3 K51

52582 52582

351

0.025%

D51

11248

2,196 104

0.15% 0.0073%

H40 D69.3

43611 8640

1,352

0.094%

D50

24947

535

0.037%

H35.3

595060

2,323

0.16%

M10, M1A

519

Lupus on Hydroxychloroquine

272

0.019%

M32

5521

Osteoarthritis on Acetaminophen Osteoporosis on Alendronate

22,395

1.6%

161

354

0.025%

M15, M16, M17, M18, M19 M80, M81

Rheumatoid Arthritis on Methotrexate ALS on Riluzole Alzheimer’s on Donepezil Bipolar Disorder on Lithium Depression on Sertraline Epilepsy on Levetiracetam Generalized Anxiety Disorder on Escitalopram Migraine on Sumatriptan

235

0.016%

M05, M06

100 362 517 4,329 1,394 619

0.007% 0.025% 0.036% 0.3% 0.097% 0.043%

G12.21 G30 F31 F32, F33 G40 F41.1

methotrexate, trexall, otrexup, rasuvo, xatmep 35623 135447 6448 36437 114477 321988

740

0.052%

G43

37418

46041

Continued on next page

Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Neurology & Psychiatry Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Oncology (Cancers) Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious Respiratory & Infectious

Myasthenia Gravis on Pyridostigmine PTSD on Sertraline Peripheral Neuropathy on Gabapentin Schizophrenia on Risperidone Acute Lymphoblastic Leukemia on Vincristine Acute Myeloid Leukemia on Cytarabine Breast Cancer on Trastuzumab Colorectal Cancer on Fluorouracil Glioblastoma on Temozolomide Lung Cancer on Pembrolizumab Melanoma on Ipilimumab Multiple Myeloma on Bortezomib Non-Hodgkin Lymphoma on Rituximab Ovarian Cancer on Carboplatin Pancreatic Cancer on Gemcitabine Prostate Cancer on Leuprolide Thyroid Cancer on Sodium Iodide I-131 Asthma on Albuterol COPD on Tiotropium Interstitial Lung Disease on Prednisone Pneumonia on Amoxicillin Sarcoidosis on Prednisone Sepsis on Vancomycin

107

0.0075%

G70.0

9000

427 1,590

0.03% 0.11%

F43.1 G62.9

36437 25480

256 110

0.018% 0.0077%

F20 C91.0

35636 11202

215

0.015%

C92.0

3041

933 1,082

0.065% 0.076%

C50 C18, C19, C20

224905 4492

103

0.0072%

C71

37776

787

0.055%

C34

1547545

244 509

0.017% 0.036%

C43 C90.0

1094833 358258

1,234

0.086%

C82, C83, C84, C85

121191

656 594

0.046% 0.042%

C56 C25

40048 12574

2,288 322

0.16% 0.023%

C61 C73

42375 9884

11,387 2,162 577

0.8% 0.15% 0.04%

J45 J44 J84

435 69120 8640

345 161 1,516

0.024% 0.011% 0.11%

J18 D86 A40, A41, O85, R65.20, R65.21, T81.12

723 8640 11124

Extended Data Table 17: Evaluation results of retrieval tasks. We report retrieval accuracy (Acc.) and mean average precision (mAP) for each of the 61 cohorts defined in Table 16. Higher (↑) values are better. Mean ± SD reported over 5 folds. Task

Model

Top-1 Acc. (↑)

Top-5 Acc. (↑)

Top-10 Acc. (↑)

mAP@5 (↑)

mAP@10 (↑)

AFib on Apixaban

Last Progress A POLLO

0.071 ± 0.001 0.302 ± 0.006

0.245 ± 0.004 0.734 ± 0.005

0.369 ± 0.002 0.870 ± 0.005

0.038 ± 0.001 0.188 ± 0.002

0.027 ± 0.001 0.151 ± 0.002

Acute MI on Aspirin

Last Progress A POLLO

0.035 ± 0.003 0.125 ± 0.006

0.120 ± 0.004 0.383 ± 0.008

0.199 ± 0.007 0.565 ± 0.013

0.016 ± 0.001 0.070 ± 0.002

0.011 ± 0.000 0.054 ± 0.002

Chronic Ischemic Heart Disease on Aspirin

Last Progress A POLLO

0.095 ± 0.001 0.339 ± 0.004

0.314 ± 0.002 0.770 ± 0.003

0.461 ± 0.003 0.889 ± 0.001

0.051 ± 0.000 0.227 ± 0.003

0.038 ± 0.000 0.190 ± 0.002

Heart Failure on Carvedilol

Last Progress A POLLO

0.024 ± 0.006 0.139 ± 0.009

0.097 ± 0.009 0.467 ± 0.014

0.162 ± 0.009 0.638 ± 0.018

0.011 ± 0.001 0.076 ± 0.003

0.007 ± 0.001 0.056 ± 0.002

Hypertension on Lisinopril

Last Progress A POLLO

0.070 ± 0.002 0.218 ± 0.003

0.276 ± 0.001 0.645 ± 0.002

0.438 ± 0.003 0.823 ± 0.003

0.035 ± 0.000 0.126 ± 0.001

0.025 ± 0.000 0.098 ± 0.001

Peripheral Artery Disease on Cilostazol

Last Progress A POLLO

0.000 ± 0.000 0.039 ± 0.011

0.025 ± 0.011 0.142 ± 0.020

0.045 ± 0.014 0.250 ± 0.020

0.002 ± 0.001 0.017 ± 0.003

0.001 ± 0.000 0.010 ± 0.002

Stroke on Alteplase

Last Progress A POLLO

0.011 ± 0.005 0.077 ± 0.013

0.054 ± 0.003 0.307 ± 0.009

0.085 ± 0.005 0.463 ± 0.013

0.005 ± 0.001 0.044 ± 0.003

0.003 ± 0.000 0.033 ± 0.002

VTE on Apixaban

Last Progress A POLLO

0.032 ± 0.005 0.127 ± 0.004

0.125 ± 0.008 0.431 ± 0.004

0.207 ± 0.006 0.626 ± 0.005

0.014 ± 0.001 0.067 ± 0.002

0.009 ± 0.001 0.048 ± 0.001

Gestational Diabetes on Insulin

Last Progress A POLLO

0.011 ± 0.005 0.140 ± 0.012

0.038 ± 0.010 0.481 ± 0.010

0.072 ± 0.014 0.627 ± 0.013

0.005 ± 0.002 0.086 ± 0.004

0.003 ± 0.001 0.064 ± 0.001

Hyperthyroidism on Methimazole

Last Progress A POLLO

0.000 ± 0.000 0.027 ± 0.009

0.003 ± 0.003 0.110 ± 0.015

0.012 ± 0.005 0.162 ± 0.013

0.000 ± 0.000 0.015 ± 0.003

0.000 ± 0.000 0.009 ± 0.002

Hypothyroidism on Levothyroxine

Last Progress A POLLO

0.051 ± 0.001 0.282 ± 0.003

0.198 ± 0.004 0.695 ± 0.004

0.331 ± 0.005 0.840 ± 0.003

0.023 ± 0.001 0.168 ± 0.001

0.016 ± 0.000 0.131 ± 0.001

Lipidemia on Atorvastatin

Last Progress A POLLO

0.106 ± 0.002 0.320 ± 0.002

0.370 ± 0.001 0.766 ± 0.003

0.545 ± 0.002 0.891 ± 0.002

0.055 ± 0.000 0.213 ± 0.001

0.039 ± 0.000 0.177 ± 0.000

PCOS on Metformin

Last Progress A POLLO

0.000 ± 0.000 0.043 ± 0.014

0.000 ± 0.000 0.186 ± 0.020

0.000 ± 0.000 0.230 ± 0.019

0.000 ± 0.000 0.020 ± 0.004

0.000 ± 0.000 0.011 ± 0.003

Preeclampsia on Magnesium Sulfate

Last Progress A POLLO

0.015 ± 0.002 0.095 ± 0.018

0.036 ± 0.006 0.318 ± 0.020

0.067 ± 0.008 0.482 ± 0.013

0.004 ± 0.000 0.047 ± 0.005

0.003 ± 0.000 0.035 ± 0.004

T2DM on Metformin

Last Progress A POLLO

0.045 ± 0.002 0.279 ± 0.004

0.167 ± 0.004 0.677 ± 0.004

0.268 ± 0.006 0.825 ± 0.007

0.020 ± 0.000 0.169 ± 0.004

0.013 ± 0.000 0.136 ± 0.003

BPH on Tamsulosin

Last Progress A POLLO

0.034 ± 0.002 0.192 ± 0.008

0.139 ± 0.006 0.578 ± 0.006

0.238 ± 0.007 0.759 ± 0.005

0.016 ± 0.001 0.106 ± 0.003

0.010 ± 0.000 0.079 ± 0.002

CKD on Lisinopril

Last Progress A POLLO

0.025 ± 0.003 0.125 ± 0.003

0.110 ± 0.003 0.438 ± 0.009

0.188 ± 0.004 0.641 ± 0.006

0.012 ± 0.001 0.065 ± 0.002

0.008 ± 0.000 0.049 ± 0.001

Cirrhosis on Propranolol

Last Progress A POLLO

0.000 ± 0.000 0.050 ± 0.013

0.028 ± 0.006 0.219 ± 0.022

0.042 ± 0.006 0.375 ± 0.033

0.002 ± 0.001 0.032 ± 0.006

0.001 ± 0.000 0.023 ± 0.003

Crohn’s Disease on Budesonide

Last Progress A POLLO

0.008 ± 0.007 0.088 ± 0.013

0.048 ± 0.013 0.304 ± 0.027

0.056 ± 0.009 0.384 ± 0.027

0.007 ± 0.003 0.051 ± 0.005

0.004 ± 0.001 0.034 ± 0.003

IBD on Mesalamine

Last Progress A POLLO

0.005 ± 0.003 0.059 ± 0.009

0.045 ± 0.006 0.193 ± 0.018

0.068 ± 0.010 0.298 ± 0.010

0.004 ± 0.001 0.025 ± 0.003

0.002 ± 0.000 0.016 ± 0.001

Ulcerative Colitis on Mesalamine

Last Progress A POLLO

0.007 ± 0.004 0.029 ± 0.008

0.033 ± 0.006 0.156 ± 0.015

0.058 ± 0.013 0.253 ± 0.031

0.004 ± 0.001 0.018 ± 0.001

0.002 ± 0.001 0.013 ± 0.001

B12 Deficiency Anemia on Cyanocobalamin

Last Progress A POLLO

0.009 ± 0.005 0.009 ± 0.005

0.023 ± 0.007 0.048 ± 0.003

0.051 ± 0.009 0.102 ± 0.010

0.005 ± 0.002 0.005 ± 0.001

0.003 ± 0.001 0.003 ± 0.001

Glaucoma on Latanoprost

Last Progress A POLLO

0.025 ± 0.004 0.097 ± 0.007

0.097 ± 0.005 0.394 ± 0.008

0.166 ± 0.007 0.578 ± 0.010

0.011 ± 0.001 0.053 ± 0.002

0.007 ± 0.000 0.039 ± 0.001

Immune Thrombocytopenia on Prednisone

Last Progress A POLLO

0.000 ± 0.000 0.038 ± 0.016

0.000 ± 0.000 0.105 ± 0.028

0.000 ± 0.000 0.143 ± 0.023

0.000 ± 0.000 0.013 ± 0.003

0.000 ± 0.000 0.007 ± 0.002

Iron Deficiency Anemia on Ferrous Sulfate

Last Progress A POLLO

0.016 ± 0.003 0.053 ± 0.005

0.056 ± 0.003 0.217 ± 0.007

0.084 ± 0.003 0.357 ± 0.007

0.006 ± 0.000 0.025 ± 0.002

0.004 ± 0.000 0.017 ± 0.001

Macular Degeneration on Ranibizumab

Last Progress A POLLO

0.189 ± 0.008 0.131 ± 0.012

0.414 ± 0.018 0.387 ± 0.007

0.498 ± 0.023 0.521 ± 0.016

0.130 ± 0.010 0.078 ± 0.005

0.103 ± 0.007 0.059 ± 0.002

Gout on Allopurinol

Last Progress

0.021 ± 0.003

0.078 ± 0.005

0.150 ± 0.004

0.008 ± 0.001

0.005 ± 0.000

Continued on next page

A POLLO

0.114 ± 0.006

0.389 ± 0.007

0.560 ± 0.004

0.062 ± 0.001

0.045 ± 0.001

Lupus on Hydroxychloroquine

Last Progress A POLLO

0.019 ± 0.005 0.106 ± 0.014

0.041 ± 0.018 0.283 ± 0.013

0.081 ± 0.015 0.386 ± 0.012

0.006 ± 0.002 0.057 ± 0.004

0.004 ± 0.001 0.040 ± 0.003

Osteoarthritis on Acetaminophen

Last Progress A POLLO

0.099 ± 0.003 0.354 ± 0.003

0.338 ± 0.002 0.767 ± 0.001

0.498 ± 0.003 0.879 ± 0.002

0.052 ± 0.001 0.242 ± 0.001

0.038 ± 0.001 0.205 ± 0.001

Osteoporosis on Alendronate

Last Progress A POLLO

0.000 ± 0.000 0.023 ± 0.012

0.006 ± 0.003 0.110 ± 0.006

0.020 ± 0.006 0.178 ± 0.013

0.000 ± 0.000 0.011 ± 0.002

0.000 ± 0.000 0.007 ± 0.001

Rheumatoid Arthritis on Methotrexate

Last Progress A POLLO

0.021 ± 0.000 0.060 ± 0.014

0.068 ± 0.013 0.166 ± 0.032

0.081 ± 0.016 0.255 ± 0.018

0.010 ± 0.001 0.021 ± 0.004

0.005 ± 0.001 0.013 ± 0.002

ALS on Riluzole

Last Progress A POLLO

0.080 ± 0.023 0.250 ± 0.032

0.160 ± 0.036 0.650 ± 0.040

0.180 ± 0.036 0.730 ± 0.023

0.020 ± 0.004 0.189 ± 0.014

0.013 ± 0.003 0.147 ± 0.010

Alzheimer’s on Donepezil

Last Progress A POLLO

0.008 ± 0.007 0.130 ± 0.010

0.033 ± 0.006 0.494 ± 0.021

0.089 ± 0.016 0.704 ± 0.018

0.004 ± 0.001 0.071 ± 0.005

0.002 ± 0.001 0.055 ± 0.002

Bipolar Disorder on Lithium

Last Progress A POLLO

0.021 ± 0.004 0.143 ± 0.015

0.052 ± 0.009 0.472 ± 0.013

0.079 ± 0.010 0.640 ± 0.006

0.006 ± 0.001 0.084 ± 0.003

0.004 ± 0.000 0.060 ± 0.002

Depression on Sertraline

Last Progress A POLLO

0.026 ± 0.001 0.078 ± 0.003

0.101 ± 0.003 0.304 ± 0.002

0.168 ± 0.003 0.482 ± 0.005

0.012 ± 0.000 0.041 ± 0.000

0.008 ± 0.000 0.028 ± 0.000

Epilepsy on Levetiracetam

Last Progress A POLLO

0.027 ± 0.005 0.151 ± 0.005

0.100 ± 0.008 0.407 ± 0.008

0.147 ± 0.006 0.550 ± 0.006

0.015 ± 0.001 0.086 ± 0.002

0.010 ± 0.001 0.066 ± 0.001

Generalized Anxiety Disorder on Escitalopram

Last Progress

0.000 ± 0.000

0.003 ± 0.002

0.010 ± 0.003

0.000 ± 0.000

A POLLO

0.013 ± 0.002

0.082 ± 0.006

0.142 ± 0.007

0.008 ± 0.001

0.005 ± 0.000

Migraine on Sumatriptan

Last Progress A POLLO

0.004 ± 0.001 0.042 ± 0.002

0.014 ± 0.003 0.160 ± 0.009

0.033 ± 0.006 0.260 ± 0.015

0.002 ± 0.000 0.020 ± 0.001

0.001 ± 0.000 0.014 ± 0.001

Myasthenia Gravis on Pyridostigmine

Last Progress A POLLO

0.019 ± 0.010 0.065 ± 0.010

0.094 ± 0.019 0.159 ± 0.028

0.122 ± 0.029 0.197 ± 0.027

0.014 ± 0.004 0.031 ± 0.007

0.010 ± 0.002 0.024 ± 0.005

PTSD on Sertraline

Last Progress A POLLO

0.000 ± 0.000 0.026 ± 0.008

0.023 ± 0.007 0.122 ± 0.007

0.035 ± 0.007 0.213 ± 0.026

0.001 ± 0.000 0.013 ± 0.001

0.001 ± 0.000 0.009 ± 0.001

Peripheral Neuropathy on Gabapentin

Last Progress A POLLO

0.009 ± 0.001 0.045 ± 0.003

0.037 ± 0.003 0.184 ± 0.009

0.085 ± 0.005 0.337 ± 0.015

0.004 ± 0.000 0.020 ± 0.001

0.003 ± 0.000 0.014 ± 0.000

Schizophrenia on Risperidone

Last Progress A POLLO

0.031 ± 0.012 0.102 ± 0.010

0.086 ± 0.020 0.305 ± 0.014

0.117 ± 0.014 0.426 ± 0.010

0.010 ± 0.003 0.041 ± 0.003

0.006 ± 0.001 0.028 ± 0.002

Acute Lymphoblastic Leukemia on Vincristine

Last Progress

0.074 ± 0.016

0.147 ± 0.015

0.184 ± 0.013

0.023 ± 0.004

0.014 ± 0.002

A POLLO

0.264 ± 0.041

0.518 ± 0.035

0.664 ± 0.028

0.143 ± 0.012

0.107 ± 0.007

Acute Myeloid Leukemia on Cytarabine

Last Progress A POLLO

0.051 ± 0.014 0.391 ± 0.021

0.220 ± 0.014 0.809 ± 0.020

0.313 ± 0.006 0.921 ± 0.011

0.026 ± 0.003 0.247 ± 0.011

0.019 ± 0.001 0.194 ± 0.008

Breast Cancer on Trastuzumab

Last Progress A POLLO

0.077 ± 0.008 0.304 ± 0.013

0.288 ± 0.006 0.753 ± 0.018

0.368 ± 0.013 0.883 ± 0.007

0.043 ± 0.003 0.181 ± 0.006

0.029 ± 0.002 0.143 ± 0.005

Colorectal Cancer on Fluorouracil

Last Progress A POLLO

0.161 ± 0.004 0.497 ± 0.013

0.349 ± 0.008 0.817 ± 0.007

0.438 ± 0.006 0.893 ± 0.006

0.086 ± 0.003 0.365 ± 0.010

0.061 ± 0.002 0.305 ± 0.008

Glioblastoma on Temozolomide

Last Progress A POLLO

0.069 ± 0.033 0.155 ± 0.040

0.195 ± 0.015 0.544 ± 0.055

0.262 ± 0.009 0.728 ± 0.029

0.030 ± 0.009 0.082 ± 0.010

0.019 ± 0.005 0.057 ± 0.005

Lung Cancer on Pembrolizumab

Last Progress A POLLO

0.094 ± 0.007 0.301 ± 0.018

0.286 ± 0.014 0.709 ± 0.013

0.394 ± 0.011 0.844 ± 0.010

0.045 ± 0.003 0.178 ± 0.004

0.032 ± 0.002 0.140 ± 0.003

Melanoma on Ipilimumab

Last Progress A POLLO

0.115 ± 0.009 0.414 ± 0.024

0.291 ± 0.011 0.742 ± 0.020

0.369 ± 0.016 0.836 ± 0.025

0.084 ± 0.007 0.273 ± 0.013

0.058 ± 0.004 0.215 ± 0.008

Multiple Myeloma on Bortezomib

Last Progress A POLLO

0.165 ± 0.021 0.475 ± 0.030

0.367 ± 0.032 0.823 ± 0.018

0.452 ± 0.034 0.894 ± 0.017

0.095 ± 0.008 0.334 ± 0.013

0.072 ± 0.005 0.279 ± 0.013

Non-Hodgkin Lymphoma on Rituximab

Last Progress A POLLO

0.140 ± 0.003 0.418 ± 0.008

0.320 ± 0.008 0.758 ± 0.010

0.409 ± 0.009 0.865 ± 0.010

0.081 ± 0.003 0.290 ± 0.008

0.060 ± 0.003 0.239 ± 0.005

Ovarian Cancer on Carboplatin

Last Progress A POLLO

0.144 ± 0.012 0.484 ± 0.023

0.316 ± 0.020 0.852 ± 0.008

0.400 ± 0.024 0.917 ± 0.011

0.087 ± 0.006 0.336 ± 0.007

0.069 ± 0.004 0.281 ± 0.005

Pancreatic Cancer on Gemcitabine

Last Progress A POLLO

0.101 ± 0.005 0.362 ± 0.019

0.267 ± 0.013 0.804 ± 0.010

0.375 ± 0.020 0.880 ± 0.012

0.047 ± 0.002 0.246 ± 0.007

0.030 ± 0.001 0.200 ± 0.006

Prostate Cancer on Leuprolide

Last Progress A POLLO

0.244 ± 0.006 0.443 ± 0.010

0.438 ± 0.009 0.787 ± 0.008

0.489 ± 0.009 0.875 ± 0.005

0.175 ± 0.002 0.318 ± 0.005

0.149 ± 0.001 0.276 ± 0.005

Thyroid Cancer on Sodium Iodide I-131

Last Progress

0.121 ± 0.019

0.282 ± 0.025

0.354 ± 0.041

0.063 ± 0.005

0.044 ± 0.005

Continued on next page

A POLLO

0.248 ± 0.021

0.540 ± 0.018

0.642 ± 0.028

0.159 ± 0.012

0.125 ± 0.009

Asthma on Albuterol

Last Progress A POLLO

0.051 ± 0.003 0.163 ± 0.004

0.188 ± 0.003 0.493 ± 0.005

0.290 ± 0.002 0.667 ± 0.004

0.026 ± 0.001 0.095 ± 0.002

0.018 ± 0.000 0.073 ± 0.001

COPD on Tiotropium

Last Progress A POLLO

0.026 ± 0.003 0.222 ± 0.009

0.116 ± 0.006 0.557 ± 0.008

0.196 ± 0.005 0.704 ± 0.009

0.012 ± 0.001 0.135 ± 0.003

0.008 ± 0.000 0.108 ± 0.002

Interstitial Lung Disease on Prednisone

Last Progress A POLLO

0.033 ± 0.006 0.151 ± 0.016

0.102 ± 0.008 0.402 ± 0.014

0.153 ± 0.005 0.523 ± 0.012

0.017 ± 0.001 0.093 ± 0.005

0.014 ± 0.001 0.068 ± 0.003

Pneumonia on Amoxicillin

Last Progress A POLLO

0.015 ± 0.004 0.072 ± 0.007

0.041 ± 0.011 0.171 ± 0.014

0.059 ± 0.011 0.252 ± 0.007

0.005 ± 0.001 0.031 ± 0.004

0.003 ± 0.001 0.023 ± 0.002

Sarcoidosis on Prednisone

Last Progress A POLLO

0.019 ± 0.007 0.019 ± 0.007

0.050 ± 0.011 0.062 ± 0.012

0.068 ± 0.013 0.149 ± 0.017

0.007 ± 0.002 0.006 ± 0.002

0.004 ± 0.001 0.005 ± 0.001

Sepsis on Vancomycin

Last Progress A POLLO

0.015 ± 0.002 0.177 ± 0.008

0.079 ± 0.003 0.543 ± 0.011

0.142 ± 0.004 0.708 ± 0.009

0.008 ± 0.001 0.096 ± 0.004

0.005 ± 0.000 0.069 ± 0.002

Record · ID 120497 · SHA-256 69f39d9ace9aa382
Retrieved via Conceptio — every document is proof-bundled with source, license, and retrieval metadata.