Proceedings of the 3rd edition of the International e-Health Forum 2024.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice BMC Proc . 2026 Mar 30;20(Suppl 11):13. doi: 10.1186/s12919-026-00366-4 Search in PMC Search in PubMed View in NLM Catalog Add to search Proceedings of the 3rd edition of the International e-Health Forum 2025 Article notes Copyright and License information Conference International E-Health forum, Center for Innovation in e-Health; Mohammed V University in Rabat, Morocco. Casablanca, Morocco 25-27 November 2025 Collection date 2026. © The Author(s) 2026, modified publication 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13034601  PMID: 41906146 This article has been corrected. See BMC Proc. 2026 Apr 15;20:20 . List of editors: Pr. Anass Doukkali, Pr. Youns Bjijou, Pr. Mohamed Adnaoui Reviewing Committee: Pr. Mohamed Adnaoui, Pr. Saber Boutayeb, Pr. Sihame Lkhoyaali, Pr. Hanaa Hachimi, Pr. Chakib El Mokhi, Pr. Laurent Meriade, Pr. Abha Cherkani Hassani, Adam Skali Correspondence: [email protected]. I1 Introduction to the 3rd edition of the International e-Health Forum – IeHF2025 A. Doukkali The e-Health Innovation Center (CIeS), Mohammed V University in Rabat Correspondence: A. Doukkali ([email protected]) BMC Proceedings 2026 , 20(11): I1 The original version of this article was revised: The article title has been corrected. This volume brings together the scientific abstracts and posters presented during the third edition of the International e-Health Forum (IeHF2025), which took place from 25 to 27 November 2025 at the Mohammed VI University of Health Sciences (UM6SS) in Casablanca. Held under the High Patronage of His Majesty King Mohammed VI, the event was co-organized by the Mohammed VI Foundation for Health and the e-Health Innovation Center (CIeS) of Mohammed V University in Rabat, with the support of various ministries and national institutions. IeHF2025 convened a wide range of national and international stakeholders committed to advancing digital health transformation across Africa and the Global South. The scientific communications included in this volume reflect the broader themes addressed during the forum, particularly the governance and interoperability of health data systems, the responsible integration of artificial intelligence in clinical care, and the use of digital technologies to support major events such as the 2030 FIFA World Cup. In addition to the scientific program, IeHF2025 featured several key initiatives that contributed to shaping the national and regional digital health agenda. These included the inaugural Health Connect Showcase on Interoperability, the first meeting of the MOHIM* Interoperability Working Group, and live demonstrations of AI solutions applied in real-world clinical settings. The MAHIR** Framework was also presented as a strategic tool to support AI adoption in health institutions. A particular emphasis was placed on digital health applications in sports medicine, as well as on fostering innovation ecosystems through the International Startup Call, the Hackathon, and the Best Abstract Awards. With a strong focus on co-creation, public-private partnerships, and regional cooperation, the final declaration of IeHF2025 outlined a shared commitment to building a robust national governance framework, advancing structured AI deployment, harmonizing interoperability standards, investing in digital skills, and reinforcing international collaboration. These efforts aim to position Morocco as a leader in secure, equitable, and innovation-driven digital health *Morocco Health Interoperability and Maturity Program **Morocco AI for Health Implementation and Readiness Framework O1 Automated electrocardiogram analysis using temporal convolutional networks A. Elmassaoudi 1 , M. Cherti 2 , S. Douzi 2 , M. Abik 1 1 National School of Computer Science and Systems Analysis (ENSIAS), Mohammed V University, Rabat, Morocco; 2 Faculty of Medicine and Pharmacy, Mohammed V University, Rabat, Morocco BMC Proceedings 2026 , 20(11): O1 Abstract Background Electrocardiogram (ECG) interpretation remains a time-consuming task and is subject to inter-observer variability, particularly in high-volume clinical environments. Cardiology departments increasingly require rapid and consistent automated ECG analysis solutions to support clinical decision-making and improve workflow efficiency. Materials and Methods A retrospective dataset of 14,827 twelve-lead ECG recordings was analyzed, with data split into training (80%), validation (10%), and test (10%) sets. Signal preprocessing included band-pass filtering between 0.5 and 45 Hz, baseline wander removal, and adaptive R-peak detection using the Pan–Tompkins algorithm. For each ECG record, nine averaged fiducial features were extracted to capture temporal and amplitude characteristics, including RR, PR, QT, ST, PQ, and QRS intervals, as well as R, Q, and P wave amplitudes. Class imbalance was mitigated by down-sampling majority normal recordings and augmenting minority classes using a one-dimensional generative adversarial network (1D-GAN). The predictive model employed a Temporal Convolutional Network (TCN) architecture composed of three dilated residual blocks, combined with a multilayer perceptron for fiducial feature integration. Training was performed using AdamW optimization with OneCycle learning rate scheduling, gradient clipping, and early stopping. Results The proposed model achieved an overall classification accuracy of 77% and a macro-averaged F 1 score of 0.70. One-vs-rest area under the curve (AUC) analysis demonstrated strong discriminative performance across cardiac conditions, including conduction disorders (0.892), hypertrophy (0.911), prior myocardial infarction (0.935), normal ECGs (0.948), and ST-T changes (0.909). Recall was highest for normal ECGs (0.87) and lowest for hypertrophy (0.51), reflecting sensitivity challenges in under-represented classes. Average inference time was below 50 milliseconds per ECG sample. Conclusions This lightweight Temporal Convolutional Network pipeline enables rapid and consistent multi-class ECG classification using band-passed signals and a limited set of fiducial features. The architecture is well suited for real-world cardiology deployment, balancing computational efficiency with diagnostic performance across multiple cardiac conditions. Future work will focus on integrating richer heart rate variability and frequency-domain features, improving data augmentation strategies for rare classes, and extending the system toward mobile and point-of-care applications to enhance robustness and generalizability. Keywords Cardiovascular diseases, electrocardiography, diagnosis, machine learning, deep learning O2 Artificial neural network-based prediction of immunotherapy response in glioblastoma patients using transcriptomic data Z. El Moudden 1 , K. Elazhary 1 , S. Souat 1 , M. Jebbar 2 , A. Badou 1 1 Immuno-Genetics and Human Pathology Laboratory, Faculty of Medicine and Pharmacy, Casablanca, Morocco; 2 Computer Science and Smart Systems (C3S) Laboratory, Higher School of Technology of Casablanca, Morocco Correspondence: A. Badou BMC Proceedings 2026 , 20(11): O2 Abstract Background Glioblastoma (GBM) is the most aggressive primary brain tumor, characterized by poor prognosis and limited response to immune checkpoint inhibitors such as anti-PD1. Predicting immunotherapy response represents a critical challenge in the field of precision oncology. Transcriptomic profiling can reveal molecular signatures of resistance, while artificial neural networks (ANNs) provide powerful tools to model complex biological datasets and improve patient stratification. Materials and Methods We analyzed transcriptomic profiles from a cohort of 34 GBM patients treated with anti-PD1 immunotherapy. Differential expression analysis was performed on 56,279 genes, identifying 278 genes upregulated in non-responders. A Kaplan-Meier survival analysis further revealed 23 genes whose elevated expression was significantly associated with poor survival. These genes were used as input features to train an ANN model based on a multilayer perceptron architecture. Data were divided into training (n = 16) and internal validation (n = 4) sets, with an independent external test set (n = 14) for final evaluation. Model performance was assessed using accuracy and mean squared error (MSE) metrics. Results The ANN model achieved excellent predictive performance, with 100% accuracy in training and internal validation, accompanied by a progressive decrease and stabilization of MSE, indicating effective learning without overfitting. External testing confirmed model robustness, yielding an accuracy of 71.43% with low error rates. These findings demonstrate the potential of ANN models to integrate transcriptomic data and accurately predict resistance to anti-PD1 therapy in GBM patients. Conclusions ANN-based predictive modeling using transcriptomic signatures can reliably stratify GBM patients according to their likelihood of response to immunotherapy. By identifying non-responders, such approaches may optimize therapeutic decisions, reduce exposure to ineffective treatments, and guide the development of personalized immunotherapeutic strategies. Validation in larger, multicenter cohorts is warranted to establish clinical utility. Keywords: Glioblastoma, immunotherapy, artificial neural networks, transcriptomics, prediction, resistance A1 Industry 4.0 for public health protection: a digital twin approach to hospital wastewater management S. Embarki 1 , Y. El Kihel 2 , B. El Kihel 1 1 Laboratory of Industrial Engineering and Seismic Engineering, Mohammed First University, Oujda, Morocco; 2 LINEACT-CESI, Bordeaux, France BMC Proceedings 2026 , 20(11): A1 Abstract Background Hospitals and healthcare facilities generate complex wastewater streams containing high organic loads, pathogenic microorganisms, pharmaceutical residues, disinfectants, and trace heavy metals. When inadequately treated, these effluents can contribute to the spread of antimicrobial resistance, contaminate surface and groundwater resources, and pose significant risks to public health. Ensuring continuous and high-quality treatment of hospital wastewater is therefore essential. Industry 4.0 technologies, particularly digital twins and predictive maintenance, offer new opportunities to enhance the reliability and performance of advanced wastewater treatment systems while protecting both the environment and surrounding communities. Materials and Methods A laboratory-scale pilot bench was developed to replicate a hospital wastewater treatment process integrating sedimentation, membrane filtration, and ultraviolet (UV) disinfection. The system was supplied with synthetic wastewater formulated to simulate typical hospital effluent, including organic matter, pharmaceutical tracers, and microbial contaminants. Internet-of-Things (IoT) sensors continuously monitored flow rate, turbidity, pH, electrical conductivity, and mechanical vibration, transmitting real-time data to a cloud-based platform. A digital twin of the filtration system was constructed to mirror the physical behavior of the treatment unit and to integrate historical and real-time operational data. Machine-learning models, including random forest and long short-term memory (LSTM) networks, were applied to multivariate data streams to predict membrane fouling, pump degradation, and UV-lamp failure, enabling predictive maintenance alerts and optimized intervention schedules. Results During preliminary three-month experimental trials, the digital twin framework predicted membrane fouling events with an accuracy of 91% and reduced unplanned system downtime by 35% compared with conventional reactive maintenance strategies. Proactive interventions maintained effluent turbidity below 1 NTU and ensured pharmaceutical compound removal rates exceeding 95%, consistently meeting stringent hospital wastewater discharge standards. The proposed architecture demonstrated scalability and interoperability with hospital information systems, supporting potential full-scale implementation. Conclusions The application of Industry 4.0 principles to hospital wastewater treatment can substantially improve system resilience, operational efficiency, and environmental protection. The proposed digital twin–based predictive maintenance approach minimizes unexpected failures, reduces the release of pathogens and pharmaceutical residues, and lowers operational costs. Future work will focus on field deployment in Moroccan hospital settings, integration with broader digital health and infrastructure platforms, and long-term evaluation of environmental, public health, and economic impacts. Keywords Hospital wastewater, digital twin, predictive maintenance, Industry 4.0, Internet of Things, healthcare infrastructure O3 Machine learning and longitudinal data for breast cancer recurrence prediction: addressing data scarcity using synthetic data generation I. Chitaouy, M. Haddouchi†, A. Berrado† AMIPS Research Team, École Mohammadia d’Ingénieurs (EMI), Mohammed V University, Rabat, Morocco Correspondence: I. Chitaouy BMC Proceedings 2026 , 20(11): O3 M. Haddouchi and A. Berrado contributed equally to this work. Abstract Background Machine learning approaches that leverage longitudinal clinical data offer significant potential for identifying breast cancer patients at elevated risk of recurrence, enabling earlier intervention and improved survival outcomes. Longitudinal datasets, which include repeated imaging, evolving biomarkers, electronic health records, and dynamic treatment responses, capture disease progression more effectively than static baseline features. However, recurrence prediction remains challenging due to limited data availability, severe class imbalance, and underrepresentation of aggressive cancer subtypes and minority populations. These constraints significantly hinder the development of robust and equitable predictive models. Materials and Methods A structured methodological review was conducted to assess synthetic data generation techniques across survival analysis, time-series modeling, and joint longitudinal–survival frameworks. Each method was evaluated against four critical criteria: the ability to handle mixed variable types (categorical and continuous), preservation of longitudinal correlations, accommodation of censored survival outcomes, and mitigation of severe class imbalance These criteria were examined in the context of real-world breast cancer datasets, including I-SPY1, METABRIC, and institutional clinical registries. These datasets highlight practical challenges such as irregular follow-up intervals, multimodal data integration (imaging, biomarkers, and EHR), and heterogeneous temporal resolution. Emphasis was placed on defining clinically grounded principles to ensure that synthetic patient trajectories remain biologically plausible, preserve covariate dependencies, and accurately reflect censoring mechanisms. Results No existing synthetic data generation approach was found to satisfy all four evaluation criteria simultaneously. Methods designed for mixed data types, such as SMOTE-NC, manage categorical and continuous variables but disrupt temporal dependencies, sometimes producing implausible tumor evolution patterns. Time-series models preserve sequential structure but struggle with mixed variable types and time-to-event outcomes. Joint longitudinal–survival models integrate repeated measurements with survival endpoints but do not adequately address class imbalance. Survival-oriented approaches typically ignore longitudinal feature dynamics altogether. These limitations reflect not only technical constraints but also a broader misalignment between current machine learning methodologies and the complexity of clinical oncology data. Conclusions Overcoming data scarcity in breast cancer recurrence prediction requires closer collaboration between machine learning researchers and clinical experts. Effective synthetic data generation must reflect real-world biological processes rather than solely reproducing statistical patterns. This is particularly critical for underrepresented patient populations, where limited data availability exacerbates inequities in predictive model performance. Advancing equity and reliability in AI-driven oncology will depend on the development of new methods that combine statistical rigor with strong clinical validity. Keywords Breast cancer prognosis, longitudinal data analysis, machine learning, cancer recurrence prediction, synthetic data generation, survival analysis, precision oncology, health data imbalance, artificial intelligence in oncology O4 Integrating serious games into continuing education for peritoneal dialysis: enhancing competency, engagement, and patient safety A. Bahadi 1,2 , M. Talaa 1 , A. Naim 1 , D. El Kabbaj 2 , M. Chahbouni 1 1 Simulation Laboratory in Health Sciences, Mohammed VI International Center for Simulation in Health Sciences, Casablanca, Morocco; 2 Mohammed V University, Rabat, Morocco BMC Proceedings 2026 , 20(11): O4 Abstract Background Peritoneal dialysis (PD) requires advanced technical skills, strict adherence to aseptic procedures, and effective patient education to prevent complications such as peritonitis. Conventional continuing education approaches for PD healthcare professionals and patients often face challenges related to low engagement and limited opportunities for safe, hands-on practice. Serious games, defined as digital games designed for educational purposes, offer an interactive, scalable, and low-risk environment to develop procedural skills, reinforce clinical decision-making, and standardize training. Materials and Methods A pilot continuing education curriculum was developed that combined a bespoke serious game focused on PD procedures and complication management with short didactic modules and structured debriefing sessions. The serious game simulated stepwise PD exchange procedures, troubleshooting scenarios such as catheter malfunction and early signs of peritonitis, and branching decision pathways supported by real-time feedback and performance metrics. Participants, consisting of nephrology trainees, completed pre- and post-intervention assessments evaluating theoretical knowledge, checklist-based procedural competence, self-reported confidence, and overall satisfaction. Engagement and perceived utility were further explored through usage analytics and qualitative feedback. Pre- and post-intervention outcomes were compared using paired statistical tests, while qualitative data were analyzed using thematic analysis. Results In this pilot study (N = 12), the integration of serious game–based training was feasible and well accepted by participants. Knowledge scores demonstrated significant improvement following the intervention. Procedural competence, assessed using standardized checklists, increased, and self-reported confidence in recognizing and managing PD-related complications improved across participants. Engagement metrics indicated high completion rates and repeated gameplay for skill reinforcement. Qualitative feedback emphasized the value of immediate feedback and increased motivation, with participants suggesting the inclusion of more complex clinical scenarios. Conclusions The integration of serious games into continuing education programs for peritoneal dialysis is feasible and associated with measurable improvements in knowledge, procedural competence, and confidence among healthcare professionals. Serious games can effectively complement traditional training approaches by providing engaging, scalable, and low-risk practice environments that may contribute to improved patient safety and standardized competency development. Larger controlled studies are needed to evaluate long-term skill retention, impact on clinical outcomes such as peritonitis rates, and overall cost-effectiveness. Keywords Peritoneal dialysis, serious games, serious gaming, continuing education, simulation-based training, patient safety, competency-based education O5 The impact of AI-based simulation on cognitive skills among healthcare students: a scoping review S. Loubbairi, O. Benbrik, H. Nassik Research and Innovation Laboratory in Health Sciences, Faculty of Medicine and Pharmacy, Ibn Zohr University, Agadir, Morocco Correspondence: S. Loubbairi BMC Proceedings 2026 , 20(11): O5 Abstract Background Cognitive skills such as critical thinking, reflective practice, clinical reasoning, and decision making are essential for safe and effective healthcare practice. Despite rapid advances in artificial intelligence (AI) and its growing integration into medical and health professions education, most existing studies have primarily focused on the development of interpersonal and communication skills. The impact of AI-based simulation on cognitive skills remains comparatively underexplored. This scoping review aimed to synthesize available evidence on the effects of AI-driven simulation on the development of cognitive skills among healthcare students. Materials and Methods A scoping review was conducted in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. Searches were performed in PubMed, Web of Science, ScienceDirect, and Scopus to identify relevant studies published between 2015 and 2025. Eligible studies included medical and nursing students who participated in AI-based simulation programs. Two independent reviewers screened s, abstracts, and full-text articles, extracted data from included studies, and grouped findings into thematic categories. Results were synthesized using a narrative approach. Results Of the 405 studies initially identified, 7 met the inclusion criteria and were included in the final review, encompassing a total of 871 healthcare students. All included studies reported a positive effect of AI-based simulation on at least one cognitive skill. Clinical reasoning was evaluated in four studies, two of which also assessed decision-making skills. Cognitive awareness, reflection, and critical thinking were each examined in only one study, with all reporting positive outcomes. Notably, none of the included studies explicitly assessed reflective practice as a distinct outcome. Regarding AI methodologies, five studies employed computational AI techniques, whereas only two relied on symbolic AI approaches. AI-based simulation tools varied widely, ranging from symbolic rule-based platforms to advanced social robots and virtual patients powered by large language models. Conclusions This scoping review indicates that AI-based simulation has a positive impact on clinical reasoning among healthcare students. However, evidence regarding its effects on decision making, critical thinking, cognitive awareness, reflection, and reflective practice remains limited and inconclusive. Computational AI approaches currently dominate the field, with symbolic AI being underutilized. Overall, AI-based simulation shows promise for fostering cognitive skills in healthcare education, but further high-quality research is needed to clarify its effectiveness across a broader range of cognitive competencies. Keywords AI-driven simulation, cognitive skills, healthcare students, critical thinking, reflective practice, clinical reasoning, decision making O6 Early detection of Alzheimer’s disease using hybrid deep learning and multi-agent systems for longitudinal modeling Y. Bouhramache 1 , K. Afdel 2 1 Laboratory of Computer Systems and Vision (LabSIV), Faculty of Sciences, Ibn Zohr University, Agadir, Morocco; 2 Department of Computer Sciences, Faculty of Sciences, Ibn Zohr University, Agadir, Morocco BMC Proceedings 2026 , 20(11): O6 Abstract Background Alzheimer’s disease (AD) is a major neurodegenerative disorder for which early detection is critical to enable timely intervention and slow disease progression. Conventional diagnostic approaches often identify AD only after substantial neuronal damage has occurred. Although artificial intelligence methods have demonstrated potential in neuroimaging-based diagnosis, many existing models fail to adequately capture longitudinal disease progression or to integrate spatial and temporal features within clinically deployable systems. Materials and Methods A multi-agent system enhanced by Model Context Protocol (MCP) servers was developed to orchestrate data preprocessing, feature extraction, and temporal modeling for early AD detection. The proposed framework integrates a hybrid deep learning architecture in which a ResNet-50 model adapted for three-dimensional magnetic resonance imaging (MRI) extracts spatial features at each time point. These features are processed through bidirectional Long Short-Term Memory (LSTM) layers to capture short- and medium-term temporal dependencies, alongside a Transformer encoder designed to model long-term disease progression patterns. A cross-attention fusion mechanism combines outputs from both temporal branches. The system was evaluated using longitudinal data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), including MRI scans and cognitive biomarkers from more than 2,000 participants classified as cognitively normal, mild cognitive impairment, or Alzheimer’s disease. Results The proposed hybrid architecture, coordinated through the MCP-based multi-agent framework, achieved a classification accuracy of 96.7%, with sensitivity of 94.2% and specificity of 98.1% for early AD detection. The model identified conversion from mild cognitive impairment to Alzheimer’s disease an average of 18 months before clinical diagnosis, providing a meaningful window for early intervention. In addition, strong performance was observed in predicting longitudinal cognitive score trajectories, with a coefficient of determination (R²) of 0.967, supporting the framework’s suitability for disease monitoring over time. Conclusions This study presents a scalable and clinically deployable framework that effectively integrates spatial and temporal modeling for early detection of Alzheimer’s disease. The combination of an LSTM–Transformer hybrid architecture with an MCP-enabled multi-agent system enables accurate prediction of disease progression and offers substantial lead time for clinical decision-making. Future work will focus on incorporating multimodal data sources and exploring federated learning strategies to improve generalizability across diverse healthcare environments. Keywords Alzheimer’s disease, early detection, longitudinal modeling, deep learning, multi-agent systems, medical imaging, disease progression prediction O7 A random PRIM-based classifier for interpretable medical diagnosis R. Nassih, A. Berrado AMIPS Research Team, École Mohammadia d’Ingénieurs (EMI), Mohammed V University, Avenue Ibn Sina, BP 765, Agdal, Rabat, Morocco BMC Proceedings 2026 , 20(11): O7 Abstract Background Decision making in healthcare requires explainable machine learning predictive models with high, precise, and actionable predictions in order to bring insights into the decision resulting from the models. To this end, we introduce in this work a Random PRIM-based Classifier (R-PRIM-Cl) framework, which relies on a transformed bump hunting algorithm, combined with metarules-based organization and systematic cross-validation to provide a rule-based classifier. The resulting rules classify each patient in a clinically meaningful subgroup and explain why they belong to a given class. Materials and Methods Five steps make up the R-PRIM-Cl framework: binary target encoding for data preparation; PRIM box construction for random feature subspace selection (peeling threshold α = 5%, pasting threshold β = 5%, minimum support 10–30%); rule conflict resolution; metarules application using the Apriori algorithm (90% confidence) for rule pruning; and a 10-fold cross-validation for final classifier selection. The framework was applied to five benchmark datasets related to breast cancer: SEER (4,024 instances, 12 attributes), ISPY1-clinical (168 instances, 18 attributes), Mammographic-masses (961 instances, 6 attributes), Wisconsin (569 instances, 32 attributes), and NKI (272 instances, 1,570 attributes), as well as the Pima Diabetes dataset (768 instances, 8 attributes). Using metrics for accuracy, precision, recall, F1-score, and ROC-AUC, performance was compared to Random Forest, XGBoost, and Logistic Regression. Results For the Diabetes dataset, R-PRIM-Cl achieved an accuracy of 95.63%, a recall of 89.43%, a precision of 92.8%, and an F1-score of 91.06%, with 93 initial rules reduced to 69 interpretable rules (maximum 4 features). The performance across the breast cancer datasets was also demonstrated: Wisconsin (accuracy 96.8%, F1-score 94.9%), SEER (accuracy 98.4%, F1-score 96.3%), ISPY1-clinical (accuracy 95.3%, F1-score 94.7%), Mammographic-masses (accuracy 97.2%, F1-score 96.1%), and NKI (accuracy 95.6%, F1-score 96.9%). R-PRIM-Cl matched or exceeded state-of-the-art algorithms while maintaining a superior precision–recall balance. ROC-AUC values exceeded 0.95 in most datasets. The framework successfully handled high-dimensional data (NKI: 1,570 features) and identified small clinically relevant subgroups (support 5–10%) overlooked by traditional methods. Metarules reduced rule redundancy by 20–40% while preserving interpretability. Conclusions Comparable to ensemble methods, R-PRIM-Cl exhibits competitive predictive accuracy and offers clear rules that clinicians can easily understand and use to prescribe treatments. Medical decision support systems, where comprehension of prediction reasoning is crucial, can benefit from the framework's special blend of prediction accuracy, interpretability, and systematic subgroup discovery. Application to a range of healthcare classification problems showcased that performance remained stable across a variety of benchmark datasets with different characteristics. O8 The role of new technologies in geriatrics and gerontology: current overview F. Boucham, R. Lemouaden, C. Elaoufir, J. Benhammou, Y. Oulehssine, A. Kadiri, A. Charef, M. Chiguer, M. Jira, F. Mekouar, N. El Omri, J. Fatihi Department of Internal Medicine B, Mohammed V Military Teaching Hospital, Rabat, Morocco BMC Proceedings 2026 , 20(11): O8 Abstract Background The natural aging of the population poses a real challenge for health systems worldwide, particularly in geriatrics and gerontology. This thesis examines the role of new technologies in the care of the elderly, focusing on their effectiveness, accessibility, and associated ethical implications. Materials and Methods A systematic literature review was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) method. Data were collected from the PubMed, Science Direct, and Google Scholar databases. The included studies were selected based on rigorous inclusion criteria, including the age of participants (65 years and older), the use of new technologies in geriatrics, and the availability of publications in English or French. Results The results show that new technologies offer numerous benefits for geriatric care. Telemedicine improves access to medical care and reduces the need for travel for consultations. Connected monitoring devices enable early detection of health anomalies, contributing to the prevention of complications. Mobile applications and smart home systems increase the autonomy and safety of the elderly. Conclusions New technologies have the power to transform geriatric care and drastically improve the quality of life for the elderly. However, ongoing efforts are necessary to overcome existing challenges and ensure the equitable and ethical adoption of these technological advances. Future research should focus on optimizing technological integration and reducing access inequalities to ensure quality care for all elderly patients. Keywords Geriatrics, gerontology, new technologies, telemedicine, gerontechnology A2 An intelligent framework for virtual synthesis and decomposition of histological skin layers I. Sehrouchni Karima 1 , L. Safae 2 , K. Salah Soumaia 2 , L. Abdelmonaime 2 1 Laboratory of Histology–Embryology–Cytogenetics, Faculty of Medicine and Pharmacy of Tangier, Morocco; 2 National School of Applied Sciences (ENSA) of Tangier, Morocco BMC Proceedings 2026 , 20(11): A2 Abstract Background Traditional histological analysis of skin tissue remains a cornerstone in the field of tissue engineering. However, it is largely qualitative and subjective, relying heavily on visual assessment by histologists. This process is time-consuming and prone to inter-observer variability. Importantly, it lacks the ability to deliver precise, quantitative insights into tissue architecture. The development of automated, objective systems to analyze and synthesize skin tissue—based on its histological features such as specific layer segmentation and matrix density—is therefore a critical unmet need. Materials and Methods We prepared a dataset of histological images through manual annotation and data augmentation. Each image contains five annotated regions representing the main skin layers: Stratum Corneum, Epidermis, Papillary Dermis, Reticular Dermis, and Hypodermis. We implemented and trained three deep learning modules, each designed for a distinct function. The Synthesis Module reconstructs virtual cross-sectional images by assembling the five individual layer images. The Decomposition Module automatically segments the five skin layers from a cross-sectional image. The Classification Module identifies and classifies each skin layer in the image. Results The Synthesis Module generated virtual cross-sections with a high mean Intersection-over-Union (IoU) of 0.94. The Decomposition Module achieved layer-wise IoU scores of: Stratum Corneum (0.96), Epidermis (0.83), Papillary Dermis (0.92), Reticular Dermis (0.96), and Hypodermis (0.96). These results demonstrate exceptional accuracy in identifying and delineating histological boundaries, closely aligning with expert histologist annotations. The Classification Module showed an overall classification accuracy of 99.65%, with precision scores of: Stratum Corneum (100%), Epidermis (99.95%), Papillary Dermis (98.61%), Reticular Dermis (99.71%), and Hypodermis (100%). In addition, the platform supports quantitative measurements of area and thickness for each layer with micrometer-level precision, enabling objective data generation that is not possible with conventional analysis methods. Conclusions This intelligent framework serves both pedagogical and diagnostic purposes. It enables the synthesis of virtual histological cross-sections from separated skin layers, providing an interactive learning tool for understanding normal and abnormal skin architecture. The Decomposition Module functions as a high-performance, objective diagnostic tool capable of segmenting skin layers while preserving their morphology and boundaries, which is particularly valuable in situations requiring precise morphometric analysis where manual or optical assessments are limited. A3 Machine learning-based prediction of coronary artery disease from ECG and TTE S. Touiti 1 , M. Hosni 2 , M. Zouga 2 , N. Fennich 1 , L. Oukerraj 1 , M. Cherti 1 1 Department of Cardiology B, Maternity Hospital Souissi, Mohammed V University, Rabat, Morocco; 2 IEST Research Team, LAIDTM, ENSAM, Moulay Ismail University of Meknes, Morocco Correspondence: S. Touiti BMC Proceedings 2026 , 20(11): A3 Abstract Background Coronary artery disease (CAD) remains the leading cause of cardiovascular mortality worldwide. Coronary angiography is the gold standard for diagnosis but is invasive, costly, and not always accessible. Artificial intelligence (AI) and machine learning (ML) may improve non-invasive prediction of coronary lesions by integrating electrocardiography (ECG) and transthoracic echocardiography (TTE). Materials and Methods This study was conducted at the Cardiology B Department, Ibn Sina University Hospital, Rabat. We retrospectively and prospectively included 140 patients admitted with suspected acute coronary syndrome who underwent ECG, TTE, and coronary angiography. Data preprocessing included imputation of missing values, standardization, one-hot encoding of coronary segments, and class balancing using SMOTE. Several machine learning algorithms were trained and optimized using nested cross-validation, including Logistic Regression, Random Forest, Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), and Naïve Bayes. Model performance was assessed by accuracy, precision, recall, and F1-score. Results In the current dataset, Random Forest achieved the best performance for predicting CAD severity, with an F1-score of 0.546, recall of 0.571, and accuracy of 0.57. SVM ranked second, with slightly higher precision but lower recall. Logistic Regression and k-NN showed intermediate results, while Naïve Bayes performed poorly. Further analyses and larger patient inclusion are ongoing, with updated results expected by November 2025 to refine predictive accuracy. Conclusions Integrating ECG and TTE data with AI and ML algorithms provides a promising non-invasive approach for early CAD detection and severity stratification. This intelligent cardiology platform may help reduce unnecessary angiographies, improve emergency triage, and support remote patient follow-up. Keywords Artificial intelligence, machine learning, coronary artery disease, electrocardiography, echocardiography, early detection A4 The effect of the use of Information and Communication Technologies (ICT) on the quality of care for tuberculosis patients in Ouarzazate B. Nadira Institut Supérieur des Professions Infirmières et Techniques de Santé d’Agadir, Agadir, Morocco BMC Proceedings 2026 , 20(11): A4 Abstract Background Tuberculosis (TB) remains one of the leading infectious causes of morbidity and mortality worldwide, particularly in low-resource settings. In Morocco, more than 30,000 new TB cases are reported annually, with a growing proportion of extrapulmonary TB. In remote provinces such as Ouarzazate, challenges in access to healthcare and treatment monitoring underscore the need for innovative solutions. Information and Communication Technologies (ICT) have the potential to enhance the quality of TB patient management by improving communication, coordination, and follow-up care. Materials and Methods This mixed-methods evaluative study was conducted at the Tuberculosis and Respiratory Diseases Diagnostic Center (CDTMR) in Ouarzazate over five months. The quantitative component involved the retrospective analysis of 455 TB patient records spanning 2020–2024. The qualitative component consisted of structured questionnaires administered to 75 healthcare professionals involved in the National TB Control Program (PNLAT) and 35 TB patients. Data analysis was performed using Epi Info 7.2.4, with descriptive and comparative statistics, and thematic content analysis for qualitative responses. Results The study population had a mean age of 43 years, with a male predominance (59%). Pulmonary TB accounted for 51% of cases, and extrapulmonary TB for 49%. Findings revealed uneven adoption of ICT tools: while most professionals recognized their value in improving communication, data management, and treatment monitoring, barriers such as limited digital infrastructure, insufficient training, and patient illiteracy reduced their effectiveness. Patients reported improved follow-up and fewer communication gaps when ICT tools such as SMS reminders, electronic drug monitoring, and teleconsultations were employed. However, 54% of patients lacked health insurance, and only 57% had access to a mobile phone, limiting widespread implementation. Conclusions ICT integration significantly enhances the quality of TB care by reducing medical errors, strengthening communication, and supporting treatment adherence. Nevertheless, structural and human barriers persist, requiring investments in digital infrastructure, professional training, and patient education. Ensuring interoperability of health information systems and adopting a participatory approach with all stakeholders are critical to scaling ICT solutions. These findings provide evidence-based insights for policymakers and health managers to leverage digital health in achieving national and global End TB targets. A5 Industry 4.0 for public health protection: a digital twin approach to hospital wastewater management S. Embarki 1 , Y. El Kihel 2 , B. El Kihel 1 1 Laboratory of Industrial Engineering and Seismic Engineering, Mohammed First University, Oujda, Morocco; 2 LINEACT-CESI, Bordeaux, France BMC Proceedings 2026 , 20(11): A5 Abstract Background Hospitals and healthcare facilities discharge complex wastewater that contains not only high organic loads but also pathogens, pharmaceutical residues, disinfectants, and trace heavy metals. If inadequately treated, these effluents can spread antimicrobial resistance, contaminate surface and groundwater, and threaten public health. Ensuring continuous, high-quality treatment of hospital wastewater is therefore critical. Industry 4.0 technologies, especially digital twins and predictive maintenance, offer new opportunities to maintain the reliability of advanced filtration systems that safeguard both the environment and surrounding communities. Materials and Methods A laboratory-scale pilot bench was designed to replicate a hospital wastewater filtration process combining sedimentation, membrane filtration, and UV disinfection. The system receives synthetic wastewater formulated to mimic typical hospital effluent, including organic matter, pharmaceutical tracers, and microbial contaminants. Internet-of-Things (IoT) sensors monitor flow rate, turbidity, pH, conductivity, and vibration, transmitting real-time data to a cloud platform. A digital twin of the filtration unit mirrors physical behavior and integrates historical data to forecast membrane fouling, pump degradation, and UV-lamp failure. Machine-learning models (random forest and LSTM) analyze multivariate data streams to generate predictive maintenance alerts and optimized intervention schedules. Results In preliminary three-month trials, the digital-twin framework predicted membrane fouling events with 91% accuracy and reduced unplanned downtime by 35% compared with conventional reactive maintenance. Early interventions preserved effluent turbidity below 1 NTU and maintained pharmaceutical removal rates above 95%, consistently meeting stringent hospital wastewater discharge standards. The architecture proved scalable and interoperable with hospital information systems for potential full-scale deployment. Conclusions Applying Industry 4.0 concepts to hospital wastewater treatment can significantly enhance system resilience and environmental protection. The proposed digital twin–based predictive maintenance strategy minimizes unexpected failures, safeguards public health by reducing pathogen and drug-residue release, and lowers operational costs. Future work will focus on field implementation in a Moroccan hospital, integration with broader digital health infrastructures, and long-term assessment of environmental and economic benefits. Keywords Hospital wastewater, digital twin, predictive maintenance, Industry 4.0, IoT, healthcare infrastructure A6 Using vibration sensors to prevent ammonia leaks in industrial refrigeration systems L. Sehli, S. Embarki, B. El Kihel Laboratory of Industrial Engineering and Seismic Engineering, Mohammed First University, Oujda, Morocco BMC Proceedings 2026 , 20(11): A6 Abstract Background Industrial ammonia-based refrigeration systems pose significant risks of hazardous leaks, primarily due to the mechanical degradation of piston compressor sealing components. Conventional preventive maintenance methods have proven insufficient for anticipating critical failures in shaft seals, piston rings, and sealing gaskets, which are responsible for approximately 60% of NH 3 leak incidents in the industry. This study investigates the predictive potential of vibration monitoring to detect early mechanical anomalies that precede ammonia leaks. Materials and Methods An integrated monitoring system was deployed on reciprocating ammonia compressors. This setup included accelerometers with real-time signal processing capabilities. The system architecture combined edge computing for local data analysis with machine learning algorithms to enable automatic fault classification. The methodology was validated over a three-month operational period, during which vibration anomalies were correlated with documented leak events to assess predictive reliability. Results Multispectral vibration analysis enabled the early detection of 80% of sealing failures, with alerts raised 24 to 72 hours before critical leaks occurred. Key vibration signatures associated with developing faults included shaft seal wear, support bearing degradation, and crack formation. When combined with existing NH 3 gas detectors, the system’s predictive reliability improved by 30% compared to setups relying on a single sensor. Implementation of this monitoring approach is projected to reduce maintenance costs by 30% and unplanned downtime by 45%, while significantly enhancing personnel safety by avoiding exposure above 10 ppm (Threshold Limit Value). Conclusions Vibration monitoring has proven to be a highly effective predictive tool for preventing ammonia leaks, supporting a transition toward Industry 4.0 maintenance strategies. The established correlation between vibration patterns and sealing system degradation paves the way for fully automated predictive diagnostics. This methodology can be extended to other high-safety, high-reliability industrial refrigeration systems. A7 KidneyLab 2.0: Arduino-based educational platform for demonstrating renal hydrosodic regulation N. E. H. Benkaddour 1 , S. Ramdani 1 , A. Messaoudi 2 , M. Boudchiche 3,4 , N. Abda 1 , Y. Bentata 1,5 1 Laboratory of Epidemiology, Clinical Research and Public Health, Faculty of Medicine and Pharmacy of Oujda, Mohammed First University, Oujda, Morocco; 2 Energy, Embedded Systems and Information Processing Laboratory, National School of Applied Sciences, Mohammed First University, Oujda, Morocco; 3 University Center for Prototyping and Innovation, Mohammed First University, Oujda, Morocco; 4 Geo-Heritage, Geo-Environment, and Mining and Water Prospecting Laboratory, Mohammed First University, Oujda, Morocco; 5 Nephrology and Kidney Transplantation Unit, Mohammed VI University Hospital, Oujda, Morocco BMC Proceedings 2026 , 20(11): A7 Abstract Background Renal hydrosodic regulation is a complex physiological process involving coordinated hormonal and cellular mechanisms essential for systemic homeostasis. Despite its significance, this subject remains challenging for students to master. Conventional teaching methods often struggle to convey the dynamic interactions underlying renal function, highlighting the need for innovative educational tools that integrate theoretical knowledge with practical experience. Materials and Methods KidneyLab 2.0 was developed as an Arduino Uno-based teaching platform designed to simulate renal hydrosodic regulation in real time. The system combines conductivity and ultrasonic sensors with the programmable Arduino Uno microcontroller, enabling measurement of sodium concentration, monitoring of fluid volumes, and reproduction of physiological regulatory responses. The platform is structured around a series of exercises, including construction of calibration curves for sodium measurement and simulation of normal and pathological states, such as hyponatremia, hypernatremia, and renal failure. Results KidneyLab 2.0 has been successfully developed and validated as a functional prototype. Using the Arduino Uno, the system reliably simulates key renal processes and produces measurable outputs that reflect changes in fluid and sodium balance under various conditions. Its modular and low-cost design facilitates reproducibility, adaptability, and potential integration into diverse teaching contexts. This approach allows demonstration of key homeostatic mechanisms, including antidiuretic hormone (ADH) activity and the renin–angiotensin–aldosterone system (RAAS). Its modular configuration permits extension to additional physiological domains, such as acid–base balance or glomerular filtration dynamics. Preliminary technical testing confirms that the platform provides accurate, real-time feedback. Conclusions KidneyLab 2.0 represents a scientifically rigorous, low-cost, and modular tool for teaching renal physiology. By translating complex regulatory mechanisms into interactive, observable simulations through an Arduino Uno-based system, the platform has strong potential to enhance understanding of hydrosodic regulation and related homeostatic processes. This initiative demonstrates the value of digitally enabled, hands-on teaching technologies in medical education and establishes a foundation for future student-based evaluation and broader application in physiology teaching. A8 Telemedicine in Southern Morocco: current state O. Benbrik 1,2 , O. Bounar 2 , I. Chakri 2 , L. Lahlou 2 1 Research and Innovation Laboratory in Health Science, Faculty of Medicine and Pharmacy, Ibn Zohr University, Agadir, Morocco; 2 Laboratory of Epidemiology, Biostatistics and Clinical Research, Faculty of Medicine and Pharmacy, Ibn Zohr University, Agadir, Morocco BMC Proceedings 2026 , 20(11): A8 Abstract Background Telemedicine refers to the delivery of healthcare services using telecommunication technologies. It has the potential to improve access to care for broader patient populations and reduce healthcare costs, particularly in resource-constrained and rural settings. Morocco has taken a leading role in the MENA (Middle East and North Africa) region by establishing a regulatory framework for telemedicine, with a national initiative launched in October 2018. As part of this initiative, Mobile Connected Medical Units (MCMUs) were deployed to address persistent gaps in healthcare coverage in medically underserved areas. This study aims to assess the current state of telemedicine implementation in Southern Morocco. Materials and Methods A descriptive secondary data analysis was conducted for the year 2024 in the Souss Massa region. Data were collected from periodic reports issued by the MCMUs operating in the region. Descriptive and comparative analyses were performed to assess usage trends, identify service gaps, and evaluate adoption patterns across different provinces. Results The telemedicine program in the Souss Massa region extended its services to three provinces: Taroudant, Tiznit, and Tata. Eight MCMUs were operational in the following locations: Tafingoult, Taliouine, Tataoute, Ait Abdell, Amelne, Reggada, Tizaght, and Irhalen. In 2024, the program served a total of 15,366 patients and facilitated 2,244 tele-expertise consultations across various medical specialties. Among these, 2,399 patients received follow-up care for diabetes, and 6,369 individuals were screened for hypertension. A positive trend was observed in the volume of health services delivered through MCMUs in 2024. The number of healthcare services and procedures increased by an average rate of 12.68% and 13.63%, respectively. Notably, tele-expertise consultations experienced a significant growth rate of 57%, reflecting a growing reliance on remote specialist support. Conclusions Telemedicine presents a significant opportunity to improve healthcare delivery and accessibility in Morocco, particularly in underserved regions. To fully realize its potential, sustained engagement of healthcare professionals and stakeholders is essential. Ensuring equitable access, strengthening infrastructure, and integrating telemedicine into broader health system planning will be key to maximizing its impact. Keywords Telemedicine; Mobile Connected Medical Units; Digital health; Morocco; Rural healthcare; Health system access A9 Can a mobile application help alleviate the workload of emergency doctors in developing countries? M. Boutkhil 1 , F.-Z. Boutkhil 2 , O. Cherradi 3 1 University College London, London, UK BMC Proceedings 2026 , 20(11): A9 Background In the intensive care unit (ICU), healthcare professionals face a high workload and must manage complex, time-sensitive tasks. Currently, many doctors rely on paper-based systems to track pending procedures, test results, and attending instructions. These paper sheets are inconvenient—they must be rewritten daily, are vulnerable to loss, and offer limited visibility into the status of tasks for other team members. Furthermore, they do not support real-time updates on task status (e.g. completed, pending, modified, or deleted), which hampers team communication and workflow coordination. Methods This study presents the design and pilot evaluation of a novel task management mobile application tailored for ICU settings in developing countries. To our knowledge, this is the first such platform developed in Morocco that aligns specifically with local ICU workflows. The application includes features such as secure login, patient case tracking, task assignment and management, and integrated team communication. Usability and performance were assessed through pilot testing involving participants simulating ICU clinical roles. Results The application supports secure clinician authentication and real-time access to patient data, enabling collaborative task execution. Key functionalities include patient case creation, efficient task delegation, and standardized instruction updates—all contributing to streamlined clinical workflows and reduced delays associated with manual record-keeping. The platform integrates role-based access control to maintain data security and define user responsibilities within care teams. Pilot testing showed promising results, although some limitations were identified, including dependency on stable internet access and an initial learning curve for users unfamiliar with digital tools. Conclusion Preliminary findings support the application’s potential to improve care delivery, reduce medical errors, and strengthen team coordination by replacing inefficient paper-based systems. The next phase will involve obtaining the necessary healthcare licensing and regulatory approvals. Following this, larger-scale clinical testing will be conducted to evaluate effectiveness, scalability, and adaptability in real-world settings. Future platform iterations will incorporate end-to-end encryption and a blockchain-based architecture to enable immutable audit trails and reinforce data integrity. A10 From coordination to explainable decision support: a roadmap for recommender systems platforms for enhanced multidisciplinary team decision-making in oncology O. El Miayar 1 , A. Berrado 1 ([email protected]) 1 Research team AMIPS, École Mohammadia d’Ingénieurs, Mohammed V University in Rabat, Avenue Ibn Sina, BP 765, Agdal, Rabat, Morocco Correspondence: O. El Miayar ([email protected]) BMC Proceedings 2026 , 20(11): A10 Background Multidisciplinary Team (MDT) meetings are a cornerstone of collaborative decision-making in oncology, bringing together clinicians from various specialties to determine optimal treatment strategies. Despite the increasing digitalisation of MDT processes globally, existing platforms often fail to combine three critical elements: interoperability, privacy safeguards, and explainable recommender systems (RS). This lack undermines trust, transparency, and clinical integration. To address these challenges, we present a roadmap for next-generation MDT platforms that go beyond coordination to include secure, explainable, RS-augmented decision support, tailored to diverse healthcare resource settings. Materials and Methods We conducted a comprehensive review of existing MDT and tumour board platforms in both scientific literature and clinical practice. This review identified major gaps in data integration, privacy protection, and explainability. From these findings, we developed a capability framework outlining five foundational pillars for responsible RS-enabled MDT adoption: (i) interoperable and high-quality data integration; (ii) documented and transparent MDT workflows; (iii) readiness for recommender systems and explainable artificial intelligence (XAI); (iv) governance and robust privacy safeguards; (v) user-friendly decision summaries for clinicians. We further propose a typology defined along four key dimensions: maturity stage, collaboration mode, AI capability, and system integration depth. Building on this, we outline a practical three-step adoption roadmap: Simulation on retrospective clinical cases; Read-only integration in real MDT meetings; Progressive operational deployment with clearly defined clinical roles, consent protocols, and audit trails. Results The proposed framework defines the minimal functional and ethical requirements for MDT platforms seeking to integrate RS and XAI features. The typology offers a shared language for stakeholders to assess, benchmark, and guide platform evolution. The adoption roadmap enables healthcare institutions—particularly those in resource-constrained settings—to experiment with RS-enhanced MDT systems with minimal disruption to existing workflows. Collectively, these contributions position RS platforms not just as technical tools, but as enablers of more accountable, efficient, and transparent group decision-making in oncology. Conclusions This roadmap provides a flexible, scalable strategy for the adoption of RS-enhanced MDT platforms, particularly suited to emerging healthcare systems such as those in Morocco. By embedding explainability, privacy, and workflow compatibility at the core of platform development, this work lays the foundation for real-world piloting, future prototyping, and rigorous clinical evaluation. Ultimately, it supports a more trustworthy and collaborative approach to oncology care. Keywords Multidisciplinary teams in oncology; tumour board; oncology decision support; recommender systems in oncology; digital health infrastructure A11 Pdaylisis – a multi-platform AI-enabled solution transforming dialysis patient care B. Abdelaali BMC Proceedings 2026 , 20(11): A11 Background Chronic kidney disease patients on dialysis face challenges including frequent hospital visits, complex treatment adherence, and limited access to personalized care. Digital health innovations can bridge these gaps by enabling real-time monitoring, predictive analytics, and patient-centered interventions. Pdaylisis is a multi-platform, multi-tenant app designed to revolutionize dialysis management by providing secure, center-specific data, role-based access for staff, and direct patient engagement. Methods Each participating center maintains an isolated database, ensuring data security and privacy. Doctors can manage patients, daily logs, treatment schedules, and full digital medical records. Patients access the platform via a unique code to view programmed treatments, medication pouch details (volume, color, and timing), and log intake/output to calculate ultrafiltration. Real-time alerts, daily indicators (weight, blood pressure, diuresis), live graphs, and two-way chat enable proactive care. A connected device under development will automate data collection. Pdaylisis is currently piloted under nephrologist Dr. Bahadi Abdelaali at Hôpital Militaire d’Instruction Mohamed V, evaluating usability, engagement, and clinical outcomes. Results Preliminary findings demonstrate increased adherence to dialysis schedules (+15%) and medication compliance (+18%). Early alerts from the AI module enabled timely interventions, preventing potential complications. Patients reported improved empowerment and ease of managing their treatment, while clinicians experienced enhanced workflow efficiency and real-time patient insights. These results highlight Pdaylisis’s capacity to improve both clinical and operational outcomes. Conclusions Pdaylisis represents a scalable, secure, and patient-centered approach to dialysis care. By integrating real-time monitoring, AI-driven alerts, and comprehensive digital records, it improves adherence, clinical oversight, and patient engagement. Future enhancements include full hemodialysis management, AI-driven decision support, voice-assisted operation, and medication stock/delivery management. Pdaylisis demonstrates a practical and innovative model for transforming dialysis care, particularly in multi-center and resource-limited settings, aligning perfectly with the goals of the International eHealth Forum 2025 to advance equitable, high-quality digital health solutions. A12 Emergency triage in the digital era: a 100% Moroccan AI chatbot proof of concept M. Choulli, M. Bahi Service des Urgences Médico-Chirurgicales, Hôpital Militaire Avicenne de Marrakech, Morocco BMC Proceedings 2026 , 20(11): A12 Background Emergency departments (EDs) in Morocco are frequently overcrowded. A significant portion of this burden comes from non-urgent cases, which consume resources meant for critically ill patients. Conversely, many patients with life-threatening conditions delay seeking care until their symptoms worsen, contributing to increased morbidity, mortality, and systemic strain. This dual challenge is further exacerbated by language barriers, as most digital health tools are available only in French or English, excluding much of the Moroccan population. In response, we developed the first 100% Moroccan, multilingual, AI-powered pre-triage chatbot accessible via WhatsApp. This tool helps patients assess the urgency of their symptoms and make informed decisions about seeking emergency care. The chatbot supports interaction in Darija, French, and English, enabling natural symptom descriptions and delivering preliminary triage guidance. Materials and Methods The chatbot prototype integrates conversational artificial intelligence, voice note processing, and image recognition capabilities. Patients can provide input through text, voice messages, or photos—for instance, images of burns, wounds, or skin lesions. Based on this information, the chatbot assigns one of three triage levels: Low urgency: consult a doctor within 24 hours Medium urgency: seek care within 6 hours High urgency: immediate referral to emergency services Results The proof-of-concept demonstrated that a multilingual AI pre-triage chatbot is feasible and relevant for the Moroccan context. Patients were able to communicate effectively in Darija, French, or English. The voice module enhanced accessibility for low-literacy users, while image analysis added diagnostic value for visible symptoms. The chatbot offers four key public health benefits: Reducing ED overcrowding by redirecting non-urgent patients to appropriate outpatient services. Encouraging timely care-seeking for urgent conditions through clear and immediate recommendations. Ensuring accessibility and inclusivity by functioning on WhatsApp and supporting local languages and cultural expressions. Providing scalability with the capacity to manage hundreds of simultaneous patient interactions without additional human resources. A prospective validation study (n=100) is currently underway at the ED of the Marrakech Military Hospital. The study compares the chatbot's triage decisions against assessments made by emergency physicians to evaluate clinical efficacy and decision accuracy. Preliminary results indicate strong potential for guiding patients toward appropriate care pathways using evidence-based logic. Conclusion This Moroccan-built AI chatbot represents a scalable, cost-effective innovation that could significantly enhance the efficiency of emergency care delivery. By helping to decongest emergency departments and promoting timely detection of critical cases, it may reduce preventable complications and improve health outcomes. Most importantly, by incorporating Darija as a core language, it addresses digital health equity in Morocco, ensuring access for all citizens, including those in rural or underserved areas. This approach aligns with national goals for inclusive, technology-enabled healthcare transformation. A13 WGS-based genomic landscape of endocannabinoid system genes in the Moroccan population: distinct novel variant identification and population-specific enrichment H. Abbou 1,2 , R. Festali 1,2 , M. W. Chemao-Elfihri 1,2 , M. Hakmi 1,2 , S. Kartti 1,2 , S. Boutayeb 1,2 , L. Belyamani 1,2,3 , R. Eljaoudi 1,4 1 Mohammed VI University of Sciences and Health (UM6SS), Casablanca, Morocco; 2 Mohammed VI Center for Research and Innovation (CM6RI), Morocco; 3 Department of Emergency, Mohammed V Military Training Hospital, Mohammed V University of Rabat, Morocco; 4 Biotechnology lab (MedBiotech), Bioinova Research Center, Medical and Pharmacy School, Mohammed V University in Rabat, Morocco BMC Proceedings 2026 , 20(11): A13 Background The endocannabinoid system (ECS), comprising cannabinoid receptors (CNR1, CNR2), metabolic enzymes (FAAH, MGLL, DAGLA, DAGLB, NAPEPLD, ABHD6, ABHD12), and auxiliary proteins (TRPV1, GPR55), regulates numerous physiological processes, including neuromodulation, pain perception, and homeostasis. While global genomic data characterize ECS variation in various populations, data specific to North Africa and Morocco remain limited. This study investigates ECS genetic diversity in Moroccans, focusing on novel variant discovery and enrichment of population-specific variants. Materials and Methods Whole-genome sequencing (WGS) data from 109 unrelated, consented Moroccan individuals were obtained via the Moroccan Genome Project. Sequencing was performed using the Illumina NovaSeq 6000 platform with 150 bp paired-end reads at a minimum depth of 30×. Multi-allelic sites were split and normalized using BCFtools v1.15.1. Variants with >10% missing genotypes, Y chromosome, and mitochondrial variants were excluded. Hardy-Weinberg equilibrium filtering (p < 5×10 -7 ) was applied via PLINK2. Variants within eleven ECS genes were annotated using Ensembl VEP v113.0, aligned to GRCh38/hg38, and incorporated ClinVar 2025 pathogenicity scores as a prioritized custom flag. Variant impact was predicted using SIFT, PolyPhen-2, and AlphaMissense. Novelty was based on the absence in global databases. Population enrichment was assessed by Fisher’s Exact Test comparing Moroccan to global allele frequencies with FDR-adjusted p < 0.05 and odds ratio >5. Results Across ECS genes, 7,415 variants were detected, predominantly modifier variants (98.6%), accompanied by low- (0.82%) and moderate-impact (0.58%) variants; no high-impact mutations were identified. We found 43 moderate-impact missense variants, including four pathogenic variants, one on CNR1 and three on MGLL. Novel variant analysis uncovered 170 previously unreported variants concentrated in MGLL, CNR2, DAGLA, and ABHD12. In parallel, enrichment analysis revealed 88 variants significantly overrepresented in the Moroccan population, especially in MGLL, ABHD12, FAAH, and DAGLB, with several variants showing odds ratios above 300. Exonic analysis highlighted Moroccan-enriched alleles in MGLL (n=227), ABHD12 (n=150), and CNR2 (n=96). Conclusions This first WGS-based characterization of ECS gene variation in Morocco reveals substantial novel and population-enriched variants with potential implications for population-specific disease susceptibility and pharmacogenomics. These findings emphasize the critical importance of including North African genomes in global datasets and set the stage for future functional and clinical investigations of ECS genetic diversity. A14 Improving access to care for children with autism in Morocco: the role and potential of telemedicine in a developing digital infrastructure C. Al Malki, M. Khalis, R. Benjelloun BMC Proceedings 2026 , 20(11): A14 Introduction Autism spectrum disorder is now a major global public health issue, affecting not only those affected but also their families, health systems, education, and society as a whole. access to care and appropriate educational interventions remains uneven across the world. School inclusion remains a challenge, particularly for children living in rural or disadvantaged areas, where the provision of specialised services is often limited (Jonge et al., 2023). However, access to specialist care remains limited, especially in rural areas (Touali et al., 2024). Moroccan families face many barriers, such as the high cost of interventions, the lack of information and the scarcity of specialised structures. Telemedicine and mobile health platforms are emerging as strategic levers in digital health infrastructure, helping to reduce disparities in access, particularly in rural or underserved areas. However, the use of these solutions to meet the specific needs of autistic children and their families remains understudied in Morocco. Materials and methods A quantitative study was conducted among 269 parents caring for children with autism in the Rabat-Salé-Kénitra, Tangier-Tetouan-Al Hoceima, and Marrakech-Safi regions. A questionnaire collected sociodemographic data on access to care, use of health services, and unmet needs. The analysis focused on the distances traveled for diagnosis, the nature of available interventions, the obstacles encountered, and the availability and potential use of digital tools such as telemedicine. Results The results show that despite the majority of families living in urban areas (76.6%), a significant number travel more than 100 km to access diagnosis. Care is mainly provided at home (92.2%), with limited but significant use of specialized interventions such as ABA, speech therapy, and psychomotor therapy. Unmet needs are high in areas such as occupational therapy (54.3%), physical therapy (34.6%), and access to a neurologist (48.4%). In addition, a significant barrier is cost (41.3%), followed by lack of information (47.2%) and waiting lists (34.6%). More than half of parents report receiving no support. These gaps reflect pressing needs that telemedicine could address by facilitating access, monitoring, and coordination of care, as well as reducing the logistical burden on families. Conclusion This study highlights that telemedicine platforms can improve access to care for children with autism in Morocco. However, their adoption requires strengthening digital infrastructure, improving knowledge, ensuring affordability, and providing psychosocial support to parents. References Jonge, M., et al. (2023). Urban-Rural Disparities in Access to Autism Services. Touali, R., Allisse, M., Zerouaoui, J., El Asri, A., El Moutawakil, B., Ouazzani, R., & Slassi, I. (2024). Anthropometric Profile, Overweight/Obesity Prevalence, and Socioeconomic Impact in Moroccan Children Aged 6–12 Years Old with Autism Spectrum Disorder. A15 Deep neural network–based generation of missing anatomy outside the field of view in computed tomography applications Y. Adib 1 , M. A. Youssoufi 2 , M. Driouch 3 , L. B. Drissi 1,4 , M. R. Mesradi 5 1 LPHE–Modeling and Simulation, Mohammed V University, Rabat, Morocco; 2 Radiotherapy Department, National Institute of Oncology, CHU Ibn Sina, Mohammed V University, Rabat, Morocco; 3 Radiotherapy Department, Clinique Spécialisée Ibn Sina, Kénitra, Morocco; 4 Hassan II Academy of Sciences and Technology, Rabat, Morocco; 5 Laboratory of Sciences and Health Technologies, High Institute of Health Sciences, Hassan First University in Settat, Settat, Morocco Correspondence: Y. Adib ([email protected]) BMC Proceedings 2026 , 20(11): A15 Abstract Background Missing anatomy—also referred to as truncation artifact—is a recurrent issue in computed tomography (CT) when the scanned anatomical region exceeds the system’s field of view (FOV). This limitation is particularly problematic in contexts such as radiotherapy planning, where precise anatomical representation is critical. This study addresses the truncation problem by introducing a deep generative neural network capable of reconstructing the missing anatomy beyond the FOV. Materials and Methods A generative model based on unsupervised deep learning was developed to reconstruct truncated CT images. The model was trained using a dataset of 25,000 lung CT images, divided into two subsets: one containing input images with artificially simulated truncation (15% to 35% of image width), and the other containing the corresponding complete (untruncated) CT images as ground truth. Data augmentation techniques included random cropping and flipping, while all voxels outside the patient’s body were assigned a fixed value of 1000 Hounsfield Units (HU) during pre-processing. The model was trained over 500 epochs (≈15 days total, ≈43 minutes per epoch). After training, the performance was validated using truncated images from 10 real patient scans. Evaluation metrics included Root Mean Squared Error (RMSE), Structural Similarity Index (SSIM), and Peak Signal-to-Noise Ratio (PSNR), calculated across training epochs and for different truncation severity levels. Results Training progression led to consistent improvements in all three metrics. For moderate truncation magnitudes (15%–23%), RMSE decreased from 5.242 to 2.456, SSIM increased from 85.7% to 95%, and PSNR improved from 19.322 to 24.664. For higher truncation levels (25%–35%), RMSE improved from 5.309 to 2.449, SSIM from 82.6% to 94.4%, and PSNR from 17.25 to 25.278. These results confirm the model’s ability to effectively reconstruct missing anatomy, with only minor performance degradation as truncation severity increases. Conclusion Using a smaller CT field of view remains a valuable strategy for reducing patient radiation exposure. However, it risks omitting essential anatomical structures—particularly in precision-critical domains like radiotherapy. This study demonstrates that generative AI (GenAI) models can effectively reconstruct missing anatomy, enabling dose reduction without compromising diagnostic or therapeutic utility. The proposed solution offers a promising approach for integrating dose optimization and image completeness in clinical CT applications. Keywords Generative Neural Network; Computed Tomography; Field-of-View; Dose Reduction; Truncation Artifact; Missing Anatomy A16 Novel 2-quinolone-triazole-α-aminophosphonate hybrids as potential chikungunya virus inhibitors K. El Gadali 1,2 , H. B. Lazrek 2 1 Laboratory of Sustainable Development and Health Research, Faculty of Sciences and Technology, Marrakech, Morocco; 2 Laboratory of Molecular Chemistry, Faculty of Sciences Semlalia, Cadi Ayyad University, Marrakesh, Morocco Correspondence: K. El Gadali ([email protected]) BMC Proceedings 2026 , 20(11): A16 Background Chikungunya virus (CHIKV), a mosquito-borne RNA virus of the Togaviridae family, causes an illness characterized by high fever and severe joint pain that can persist for extended periods. Since its global re-emergence in 2005, CHIKV has been reported in more than 100 countries, with over two million cases of acute and chronic arthritis. In the absence of widely available vaccines or targeted antiviral therapies, treatment of chronic CHIKV infection remains symptomatic, relying on immunomodulatory agents. α-Aminophosphonates have attracted significant interest due to their structural similarity to α-amino acids, peptides, and natural phosphates, as well as their broad biological activity profiles. Materials and Methods We synthesized a library of eighteen hybrid molecules incorporating three pharmacophores: 4-methyl-2-quinolone, 1,2,3-triazole, and α-aminophosphonate. The hybrids were constructed via copper-catalyzed 1,3-dipolar cycloaddition to generate the quinolone-triazole scaffold, followed by a Kabachnik–Fields reaction to append the α-aminophosphonate unit. These compounds were evaluated in vitro for anti-CHIKV activity using a cytopathic effect (CPE)-based assay on Vero A cells. Antiviral efficacy and cytotoxicity were determined by measuring IC 50 (concentration inhibiting 50% of CPE) and CC 50 (concentration reducing cell viability by 50%), respectively. Results Seven of the synthesized compounds exhibited moderate to strong anti-CHIKV activity. Two derivatives were identified as particularly potent, with IC 50 values of 9.40 μM and 6.80 μM, both surpassing the activity of the reference drug chloroquine (IC 50 = 11 μM). The CC 50 values for these compounds were 23.41 μM and 25.78 μM, resulting in selectivity indices (SI) of 2.49 and 3.79, respectively. Molecular docking studies targeting the CHIKV nsP3 macrodomain (PDB ID: 6VUQ) showed that these compounds bind with high affinity, forming multiple stabilizing interactions such as hydrogen bonds and hydrophobic contacts. Conclusion The results support the potential of quinolone-triazole-α-aminophosphonate hybrids as effective inhibitors of CHIKV replication, likely through disruption of the nsP3 macrodomain function. These findings establish a promising structural framework for the development of novel, targeted antiviral agents against chikungunya virus. Keywords Chikungunya virus; α-aminophosphonates; 1,2,3-triazole; 2-quinolone; antiviral agents; nsP3 macrodomain; structure-based design; Kabachnik–Fields reaction References Weber, W.C., Streblow, D.N., Coffey, L.L. Chikungunya Virus Vaccines: A Review of IXCHIQ and PXVX0317 from Pre-Clinical Evaluation to Licensure, BioDrugs , 38 (2024), 727–742. 10.1007/s40259-024-00677-y Hartwich, A., Zdzienicka, N., Schols, D., Andrei, G., Snoeck, R., Głowacka, I.E. Design, synthesis and antiviral evaluation of novel acyclic phosphonate nucleotide analogs with triazolo[4,5-b]pyridine, imidazo[4,5-b]pyridine and imidazo[4,5-b]pyridin-2(3H)-one systems, Nucleosides, Nucleotides & Nucleic Acids , 39 (2020), 542–591. 10.1080/15257770.2019.1669046 A17 Integrative radiogenomics and artificial intelligence: redefining precision medicine through multiscale data convergence N. Messoudi 1 , M. Es-Saadi 1 , A. Maaroufi 2 , S. Hamdi 1 1 Virology and Environmental Health Laboratory, Institut Pasteur du Maroc, Casablanca, Morocco; 2 Institut Pasteur du Maroc, Casablanca, Morocco BMC Proceedings 2026 , 20(11): A17 Abstract Complex human diseases are characterized by profound biological, molecular, and phenotypic heterogeneity, making accurate diagnosis, prognosis, and therapeutic targeting particularly challenging. Although omics-based technologies have significantly deepened our understanding of disease mechanisms, they fall short of capturing the spatial and functional organization of tissues within their physiological context. Radiogenomics bridges this gap by linking quantitative imaging features with underlying molecular and cellular alterations, offering a framework to relate observable phenotypes to their biological substrates. The integration of radiogenomics with artificial intelligence (AI) enables the convergence of multiscale data—spanning medical imaging, molecular profiles, clinical records, and environmental determinants—into unified models that support advanced patient stratification and truly individualized medical care. This study explores how such convergence is redefining the foundations of precision medicine by facilitating non-invasive disease characterization, predictive modeling, and the design of adaptive therapeutic strategies. A comprehensive critical review of literature published between 2014 and 2024 was undertaken, focusing on advances in radiogenomic modeling, multi-omics integration strategies, and translational applications across diverse disease areas, molecular pathways, and clinical phenotypes. The synthesis highlights recent methodological innovations, including deep learning architectures tailored for feature extraction and pattern recognition across imaging and omic domains. The evidence reviewed demonstrates that AI-enabled radiogenomic approaches can uncover previously hidden associations between imaging features and molecular alterations, predict therapeutic responses, and model disease progression over time. These integrative tools hold significant promise for refining diagnostic pathways, optimizing treatment decisions, and reducing reliance on invasive procedures. However, persistent challenges remain in data standardization, algorithm interpretability, and clinical validation across diverse populations and healthcare systems. The convergence of radiogenomics and AI marks a paradigm shift toward integrative precision medicine, where imaging, molecular, and clinical data coalesce into a multidimensional and dynamic model of health and disease. Realizing this vision will require the establishment of standardized interoperability frameworks, the development of transparent and explainable AI models, and the strengthening of interdisciplinary collaboration among clinicians, data scientists, and systems biologists. Keywords Radiogenomics; artificial intelligence; precision medicine; multi-omics integration; deep learning; biomarkers; systems medicine A18 Artificial neural network for predicting complication types in patients with TIVAD during chemotherapy Kawtar Matrab 1 , Amine En-Naaoui 2 , Banacer Himmi 3 , Saber Boutayeb 1,2 1 Faculty of Medicine and Pharmacy, University Mohammed V, Rabat, Morocco; 2 Mohammed VI Center for Research and Innovation (CM6RI), Rabat, Morocco; 3 Higher Institute of Nursing Professions and Health Techniques (ISPITS) of Rabat, Ministry of Health and Social Protection, Morocco BMC Proceedings 2026 , 20(11): A18 Background The use of Totally Implantable Venous Access Devices (TIVAD) is a standard practice in oncology, allowing long-term venous access for chemotherapy. However, the implementation and management of TIVAD remain critical procedures impacting patient safety. Complications during treatment, particularly infections and thrombosis, are commonly reported and can lead to significant morbidity. Early prediction of the type of complication based on clinical parameters could support preventive strategies and improve patient outcomes. This study aims to develop an artificial intelligence–based model capable of predicting whether a patient will experience infection or thrombosis if a complication occurs. Materials and Methods A dataset of 723 patients who received TIVAD at the National Institute of Oncology of Rabat from January 2023 to July 2025 was collected. Among these, 89 patients (≈12%) experienced complications. Five clinical parameters were considered as inputs for the predictive model: sex, age, history of prior complications, side of TIVAD placement, and location of diagnosis. An Artificial Neural Network (ANN) was designed with five input neurons, a hidden layer of ten neurons, and two output neurons representing the complication classes. Data were randomly split into training (70%), validation (15%), and testing (15%) subsets to evaluate model performance and avoid overfitting. Results The trained model demonstrated an overall accuracy exceeding 72% across training, validation, and testing sets. Specifically, the testing subset achieved 85% accuracy in predicting the complication class. Confusion matrix analysis revealed that the model effectively differentiated between infections and thromboses, with a balanced performance across both categories. These results suggest that the ANN can provide reliable predictions based on the selected clinical parameters, despite the relatively small proportion of complicated cases. Conclusion This study presents a machine learning approach to predict potential TIVAD-related complications in oncology patients. The model shows promising predictive capabilities and could support clinical decision-making by identifying patients at higher risk of specific complications. Nevertheless, limitations must be considered. The dataset size is limited, and complications are inherently stochastic events influenced by factors beyond the collected parameters. Future work will focus on increasing the dataset, incorporating additional clinical and biological parameters, and exploring alternative machine learning architectures to enhance prediction accuracy. Overall, this approach illustrates the potential of AI-based models in improving patient safety and guiding preventive interventions in oncology practice. A19 Aesthetic rehabilitation with injected composite in a case of dental fluorosis: smile planning using exocad software B. El Hammi, I. Ihoume, H. Moussaoui, A. Bennani Department of Fixed Prosthodontics, Faculty of Dentistry, Hassan II University, Casablanca, Morocco BMC Proceedings 2026 , 20(11): A19 Introduction The smile is a fundamental element of psychosocial well-being. Its rehabilitation represents a significant challenge in aesthetic dentistry, requiring precise and predictable planning. Artificial intelligence (AI) and digital tools, such as Computer-Aided Design (CAD) software, now offer innovative solutions for treatment planning and simulation. The objective of this clinical case report is to illustrate the contribution of Digital Smile Design (DSD) using exocad software in the planning and execution of an aesthetic rehabilitation for a patient suffering from dental fluorosis. Methods A patient presenting with unaesthetic dental fluorosis in the anterior segments was treated. The methodology was based on an integrated digital approach. Smile planning was carried out using the Digital Smile Design module of exocad software. Following the acquisition of photographs and a dynamic video of the patient's smile, an aesthetic analysis and a virtual design of the new smile were performed. This digital treatment plan served as a guide for the direct composite restorations. Results Virtual planning allowed for the establishment of a precise aesthetic project, which was approved by the patient. Using the digital guide, aesthetic restorations with injected composite were performed from tooth 15 to tooth 25 (from the second right premolar to the second left premolar). The final result showed a significant improvement in smile aesthetics, with the elimination of fluorosis stains and harmonization of the shapes, proportions, and shade of the teeth. Occlusal function was preserved. Conclusion This clinical case demonstrates that the combination of Digital Smile Design with exocad software and the injected composite technique is an effective strategy for managing the aesthetic consequences of dental fluorosis. The digital planning protocol ensured predictable results, enhanced communication with the patient, and provided precise guidance during the clinical phase. This minimally invasive approach successfully restored aesthetics and function, representing a reliable conservative solution for anterior rehabilitations. The patient provided explicit and informed consent for the publication of their clinical information in an open-access, online journal A20 AI and machine learning in antimicrobial resistance: toward faster detection and smarter antibiotic use A. Er-Regragui 1,2 , M. Snoussi 1,7,8 , H. Mguild 1,3 , R. Festali 1,2 , H. Houssam 1,2 , K. Nayme 3 , N. Nzoyikorera 4 , F. Z. Benbouazza 5 , M. Kettani-Halabi 6 , A. Chakib 1,7,8 , N. Dini 1,7,8 , I. Diawara 1,2,9 1 Mohammed VI University of Health Sciences (UM6SS), Faculty of Medicine, Laboratory of Microbiology, Infectious Diseases, Allergology and Pathogen Surveillance (LARMIAS), Casablanca, Morocco; 2 Mohammed VI Center for Research and Innovation, Rabat, Morocco; 3 Molecular Bacteriology Laboratory, Pasteur Institute of Morocco, Casablanca, Morocco; 4 National Reference Laboratory, National Public Health Institute, Bujumbura, Burundi; 5 Faculty of Sciences and Technology, Hassan First University, Settat, Morocco; 6 Mohammed VI University of Health Sciences (UM6SS), Faculty of Pharmacy, Casablanca, Morocco; 7 Cheikh Khalifa International University Hospital, Casablanca, Morocco; 8 Mohammed VI International University Hospital, Bouskoura, Casablanca, Morocco; 9 UM6SS, Higher Institute of Biosciences and Biotechnologies, Casablanca, Morocco Correspondence: A. Er-Regragui ([email protected]) BMC Proceedings 2026 , 20(11): A20 Abstract Antimicrobial resistance (AMR) is a critical global health threat, contributing to increased morbidity, mortality, and healthcare costs by rendering common infections harder to treat. Traditional diagnostic methods for AMR detection are often slow, labor-intensive, and limited in sensitivity. This study presents recent developments in the use of artificial intelligence (AI) and machine learning (ML) models to accelerate AMR detection, enhance resistance monitoring, and optimize antibiotic stewardship. A systematic review was conducted across PubMed, Scopus, and Web of Science databases for studies published between 2015 and 2025. Eligible publications included original research and review articles focusing on AI-driven models for bacterial identification, resistance prediction, or support in antibiotic prescription. Search terms included “artificial intelligence,” “machine learning,” “predictive models,” and “antimicrobial resistance.” Studies lacking experimental validation, written in languages other than English or French, or limited to conference abstracts were excluded. Out of 28 initially retrieved studies, 15 met the inclusion criteria. The findings reveal that AI models—particularly those using machine learning algorithms such as random forests and support vector machines—achieve predictive accuracies exceeding 80% for various bacterial resistance profiles. Deep learning models, especially deep neural networks, demonstrated superior performance in detecting complex resistance genes directly from genomic data. However, they require large, well-curated datasets to maintain accuracy and generalizability. Several AI-powered decision support systems reviewed in the literature also showed a capacity to reduce inappropriate antibiotic prescriptions by up to 30%, highlighting their practical value in clinical settings. These results underscore the transformative potential of AI in the fight against AMR. Predictive models based on AI not only offer faster detection but also support more precise and rational antibiotic use. Nonetheless, broad clinical validation and the seamless integration of AI tools into existing healthcare infrastructures remain major challenges. AI should be regarded as a complementary asset to traditional surveillance and prevention systems, reinforcing global efforts to curb the spread of antimicrobial resistance. Keywords Antimicrobial resistance; artificial intelligence; machine learning; deep learning; predictive models; antibiotic stewardship; resistance surveillance A21 From days to hours: AI-supported LAMP assay for rapid diagnosis of invasive non-typhoidal salmonella detection A. E. Dickson 1,2,3 , C. H. Dicko 1,2 , A. Chakib 1,2,4,5 , N. Dini 1,2,4,5 , L. A. Basing 3 , I. Diawara 1,2,6 1 Research Laboratory of Microbiology, Infectious Diseases, Allergology and Pathogen Surveillance (LARMIAS), Mohammed VI Faculty of Medicine, Mohammed VI University of Sciences and Health (UM6SS), Casablanca 82403, Morocco; 2 Mohammed VI Higher Institute of Biosciences and Biotechnologies, Mohammed VI University of Sciences and Health (UM6SS), Casablanca 82403, Morocco; 3 Department of Medical Diagnostic, Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana; 4 Cheikh Khalifa International University Hospital, Casablanca 82403, Morocco; 5 Mohammed VI International University Hospital, Bouskoura 27182, Morocco; 6 Mohammed VI Center for Research and Innovation (CM6RI), Rabat 10112, Morocco BMC Proceedings 2026 , 20(11): A21 Background Invasive non-typhoidal Salmonella (iNTS) remains a significant cause of bloodstream infections in sub-Saharan Africa, where young children are particularly at risk. Immunocompromised individuals, sickle cell patients, malnourished individuals and people with malaria and other diseases are equally at risk. Routine diagnosis especially in sub–Saharan Africa depends on conventional blood culture, which is slow, requires laboratory infrastructure, and often yields low sensitivity. These limitations delay treatment and contribute to high mortality. Developing rapid, accurate diagnostic tools that can be used in resource-constrained settings is therefore essential. Methods Clinical and foodborne iNTS isolates were collected in Ghana and Morocco and subjected to culturing, identification and then whole genome sequencing to identify virulence and resistance markers suitable for diagnostic targeting. A loop-mediated isothermal amplification (LAMP) assay was then designed, with artificial intelligence applied to assist primer selection and improve amplification efficiency. The assay was validated against conventional blood culture and polymerase chain reaction, with performance measured in terms of sensitivity, specificity, and time-to-result. Results Analysis of the genomic data highlighted conserved gene regions associated with invasiveness and antimicrobial resistance. Incorporating artificial intelligence into assay design reduced development time and improved reaction reliability. In preliminary validation, the LAMP assay achieved sensitivity and specificity above 90 percent. Crucially, the turnaround time was reduced from several days with culture to under two hours, offering a major improvement in speed. The assay also proved more adaptable to limited-resource laboratory conditions than polymerase chain reaction. Conclusions This study demonstrates the potential of artificial intelligence-enhanced molecular diagnostics to close critical gaps in the detection of invasive non typhoidal salmonella. The test provides a faster and more reliable means of diagnosis, enabling earlier treatment decisions and more effective antimicrobial use. These advances could reduce childhood mortality linked to invasive salmonellosis while supporting public health efforts to contain antimicrobial resistance. A22 Role of the shared electronic health record in clinical coordination between primary-level physicians and referral-level specialists in Morocco R. Moulki 1,2,3 , Z. Belrhiti 1,2 , H. Asri 3,4 , A. El-Ammari 3 , A. Khattabi 1,2,3 1 Mohammed VI International School of Public Health, Mohammed VI University of Health Sciences (UM6SS), Casablanca, Morocco; 2 Management and Public Health Laboratory, Mohammed VI Center for Research and Innovation (CM6RI), Rabat, Morocco; 3 National School of Public Health (ENSP), Ministry of Health and Social Protection, Rabat, Morocco; 4 Regional Directorate of the Ministry of Health and Social Protection, Marrakech–Safi, Marrakech, Morocco Correspondence: R. Moulki ([email protected]) BMC Proceedings 2026 , 20(11): A22 Abstract Clinical coordination is a central objective of Morocco’s ongoing National Health System reform. Framework Law 06-22 identifies the digitalization of the health information system—particularly through the Shared Electronic Health Record (SEHR)—as a strategic tool for improving service integration across levels of care. This study aimed to assess the perceived level of clinical coordination between primary-level physicians (PLPs) and referral-level specialists (RLSs), evaluate the use and perceived utility of nine coordination mechanisms with particular focus on the SEHR, and identify barriers to its effective use. A cross-sectional survey was conducted between April and May 2024 in the Casablanca–Settat region among 329 public-sector physicians (186 PLPs and 143 RLSs). Data were collected using the COORDENA-CAT questionnaire, adapted to the Moroccan context. Variables included perceived coordination, access to and frequency of use of coordination mechanisms (daily/weekly), perceived usefulness, and barriers to SEHR implementation. Differences between PLPs and RLSs were analyzed using Chi-square tests, with significance set at p < 0.05. Findings indicate that coordination between care levels is generally perceived as insufficient, with RLSs expressing more favorable views than PLPs. The most commonly used mechanism was the Referral/Counter-Referral (RCR) system, used by 70.2% of respondents (PLPs 91.9% vs RLSs 42.0%, p < 0.001), with 51.4% reporting weekly/daily use and 97.6% considering it useful. The telephone was used by 74.8% (PLPs 78.0% vs RLSs 70.6%, p = 0.129), with 50.8% using it weekly/daily and 85.1% rating it as useful. Social networks were used by 42.9% (PLPs 39.8% vs RLSs 46.9%, p = 0.199), with 29.5% using them weekly/daily and 65.3% finding them useful. The SEHR was reported as used by only 12.5% of respondents (PLPs 7.0% vs RLSs 19.6%, p = 0.001), with weekly/daily usage at just 9.7%. Despite this low adoption rate, its perceived usefulness was high (89.4%). Reported barriers to SEHR use included technical limitations (lack of interoperability, low digital literacy), infrastructure deficiencies (equipment and connectivity), organizational challenges (poor integration into clinical workflows), and concerns over data confidentiality and security. These results reveal significant disparities between PLPs and RLSs in perceived coordination, particularly concerning the SEHR. While traditional tools such as RCR, telephone, and social media remain in widespread use, they lack traceability, standardization, and data protection. Conversely, the SEHR, although underused, is conceptually well-accepted and aligns closely with national policy directions under Framework Law 06-22. To harness its full potential, substantial investments are required in infrastructure, digital capacity-building, organizational integration, and data governance. The SEHR thus stands as a key strategic instrument for reducing systemic asymmetries and strengthening coordination between frontline and specialist care. Keywords Clinical coordination; Shared Electronic Health Record; Coordination mechanisms; Digital health; Health system reform in Morocco A23 Application of the accuracy profile as a chemometric tool for validation of a chromatographic method for quantification of several prohibited compounds in equine anti-doping control W. El-Ghaly 1 , T. El Kamli 2 , L. Zaari Lambarki 3 , A. Benmoussa 1 , F. Bakkali 1 , T. Saffaj 3 , F. Jhilal 4 1 Mohammed VI University of Sciences and Health, Drug Sciences Laboratory, Faculty of Pharmacy, Casablanca 82403, Morocco; 2 Hassan II Agronomic and Veterinary Institute, Department of Veterinary Biological Sciences and Pharmaceuticals, Rabat 10101, Morocco; 3 Sidi Mohamed Ben Abdallah University, Applied Organic Chemistry Laboratory, Faculty of Sciences and Technology (FST), B.P. 2202, Fes 30000, Morocco; 4 Bishop’s University, Department of Chemistry and Brewing Science, Sherbrooke, QC J1M 1Z7, Canada BMC Proceedings 2026 , 20(11): A23 Abstract Background Equine doping represents a major challenge in competitive sports, undermining animal welfare and the integrity of events. In the absence of a universal quantification method, laboratories are required to develop and validate their own analytical procedure to detect and quantify doping substances, which are of particular concern due to their frequent use and potential impact on horse performance. Materials and Methods The aim of this study was to validate a bioanalytical method for the quantification of several prohibited substances in equine anti-doping control, which were tested, spiked with known concentrations of the target compounds, and analyzed to evaluate the method’s performance using ultra-high-performance liquid chromatography coupled with high-resolution mass spectrometry (UHPLC/HRMS). Validation followed international guidelines, especially the SFTSP commission, and included assessment of specificity, linearity, limit of detection (LOD), limit of quantification (LOQ), trueness, and precision [1]. Results The validated method demonstrated satisfactory linearity for all tested molecules. Achieved LOD and LOQ values complied with regulatory requirements. The intermediate precision tested under different days was satisfactory (CV < 15%). Specificity confirmed clear differentiation between target analytes and potential interferences in the horse urine matrix, while the accuracy profile—a simple and graphical decision tool using the notion of total error [2]—showed β > 0.80, indicating that at least 80% of the future results were included in the predefined range of acceptability. Conclusion This study confirms the reliability of the developed method, and the overall results confirm that it is performant in terms of selectivity, linearity, accuracy, precision, and trueness and can be easily used to ensure the credibility of equine anti-doping testing. Future perspectives include extending the method to additional substances from different classes. References El-Ghaly W, El Kamli T, Gongbe AMA, Zaari Lambarki L, El Hamdani M, Lahkak FE, Al Idrissi N, Benmoussa A, Balouch L, Bakkali F, Saffaj T, Jhilal F. Development and validation of a quantitative UHPLC-HRMS bioanalytical method for equine anti-doping control. J Pharmacol Toxicol Methods. 2025;134:107759. 10.1016/j.vascn.2025.107759 Hubert P, Nguyen-Huu JJ, Boulanger B, Chapuzet E, Chiap P, Cohen N, Compagnon PA, Dewé W, Feinberg M, Lallier M, Laurentie M, Mercier N, Muzard G, Nivet C, Valat L. Harmonization of strategies for the validation of quantitative analytical procedures. J Pharm Biomed Anal . 2004;36(3):579–586. 10.1016/j.jpba.2004.07.027 A24 The European health data space: architecture, governance, and implications for cross-border care and health data reuse A. Skali Institute for Human-Centered Health Innovation – IHCHI, Basel, Switzerland BMC Proceedings 2026 , 20(11): A24 Background The European Health Data Space (EHDS) is the European Union’s flagship regulatory initiative for enabling secure primary and secondary use of health data across Member States. Anchored in legislation such as the GDPR, Data Governance Act, Data Act, eIDAS, and the EU Cybersecurity Act, the EHDS introduces harmonised rules, cross-border data services (MyHealth@EU for primary use; HealthData@EU for secondary use), and national Health Data Access Bodies (HDABs). The policy aims are twofold: to enhance the continuity and quality of care by enabling patient data portability, and to accelerate research, regulation, public health, and innovation by facilitating trusted data reuse. Methodology This review draws from: (i) non-structured interviews with experts in public administration, healthcare provision, and digital health innovation; (ii) analysis of EU regulatory proposals and legislative texts supporting EHDS deployment; and (iii) examination of gray literature, including technical documentation, national implementation plans, and cross-border programme roadmaps. The objective is to provide a clear synthesis of EHDS design, functionality, and implications for European healthcare and innovation ecosystems. Results The analysis clarifies the EHDS architecture: mandatory EU-wide formats and interoperability rules for priority electronic health record (EHR) categories (e.g., summaries, prescriptions, imaging, lab results, discharge notes); patient control over data access; and governed secondary use through HDAB-issued permits within secure processing environments. Expected outcomes include better cross-border care, reduced diagnostic duplication, improved pharmacovigilance and surveillance, and streamlined multi-country research. Core enablers are technical standardisation, semantic and procedural interoperability, consent frameworks, transparency mechanisms, and data quality labeling. Challenges include national disparities in digital readiness, inconsistent EHR definitions, fragmented legacy systems, limited capacity for data stewardship, unclear liability for patient-generated data, and regulatory fragmentation hindering startup scaling. Case studies show centralised coordination accelerates interoperability, although adaptable models are needed for large or federal systems. Conclusion The EHDS is positioned to become core infrastructure for European digital health, provided implementation aligns governance, interoperability, consent, and security with practical workflows and capacities. Successful roll-out depends on phased obligations, robust HDAB operations, high-quality data pipelines, and inclusive stakeholder engagement. If these conditions are met, EHDS can strengthen care continuity, enable trustworthy data reuse, and catalyse innovation (including AI) at European scale. Keywords European Health Data Space; digital health; interoperability; secondary use; Health Data Access Bodies; MyHealth@EU; HealthData@EU; GDPR; Data Governance Act; Data Act; consent management; data quality A25 Digital technologies in medical mycology: MALDI-TOF mass spectrometry and automated systems for rapid identification of Candida species B. Jabri 1,2 , F. El Falaki 1 , F. Assi 3,4 , S. Faouzi 2 , A. Hbibi 5 , M. Achmit 6,7 1 Research Laboratory: Care, Health and Environment, High Institute of Nursing Professions and Health Technics, Rabat, Morocco; 2 Research Laboratory in Oral Biology and Biotechnology, Faculty of Dental Medicine, Mohammed V University in Rabat, Morocco; 3 Faculty of Medicine and Pharmacy, Mohammed V University in Rabat, Morocco; 4 Central Laboratory of Parasitology and Mycology- Ibn Sina University Hospital of Rabat, Morocco; 5 Department of Periodontology, International Faculty of dental Medicine, College of Health Sciences, International University of Rabat, Morocco; 6 High Institute of Nursing Professions and Health Technics, Casablanca, Morocco; 7 UR Microbiology, Biomolecules and Biotechnology Laboratory of Physical Chemistry and Biotechnologies of Biomolecules and Materials FST Mohammedia, Morocco Correspondence: B. Jabri ([email protected]) BMC Proceedings 2026 , 20(11): A25 Abstract Background Rapid and accurate identification of Candida species is critical in medical microbiology, given the rise of non-albicans species with diverse antifungal resistance profiles. Advances in digital technologies such as matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and automated biochemical systems have transformed laboratory workflows, enabling integration with laboratory information systems (LIS) and bioinformatics pipelines for real-time reporting. This study aimed to evaluate and compare the diagnostic performance of MALDI-TOF MS, automated systems, rapid tests, and classical methods against multiplex PCR, emphasizing their role in the digital transformation of microbiological diagnostics. Methods A total of 273 clinical Candida isolates were analyzed in two academic laboratories. Identification methods included MALDI-TOF MS (Bruker Microflex LT), VITEK 2 YST automated system, API 20C AUX biochemical profiling, rapid antigen detection tests (Bichro-Latex, Bichro-Dubli, Krusei-color), and the germ tube test. Multiplex PCR targeting six medically important Candida species served as the reference method. Diagnostic accuracy was assessed through sensitivity, specificity, predictive values, and Cohen’s Kappa coefficient. Results MALDI-TOF MS achieved almost perfect agreement with PCR (Kappa = 0.986), correctly identifying all C. albicans, C. glabrata, C. parapsilosis, and C. tropicalis isolates, with minimal discrepancies for C. dubliniensis and C. krusei. VITEK 2 YST also demonstrated excellent performance (Kappa = 0.964). API 20C AUX showed good agreement (Kappa = 0.933) but produced minor misidentifications for C. tropicalis and C. parapsilosis. Rapid tests exhibited variable results: the combined Bichro test showed high accuracy for C. albicans (Kappa = 0.899), while Bichro-Dubli identified C. dubliniensis reliably (Kappa = 0.782). The germ tube and Krusei-color tests had lower sensitivity and agreement. Conclusion MALDI-TOF MS and automated systems represent high-performance digital solutions for microbiological diagnostics, offering speed, accuracy, and seamless integration into LIS and bioinformatics workflows. Their implementation can reduce turnaround times, improve therapeutic decision-making, and strengthen laboratory interoperability within e-health ecosystems. Rapid antigen tests remain valuable in low-resource contexts when integrated into hierarchical diagnostic strategies. The findings support further adoption of digital diagnostic technologies in medical mycology and their integration into national e-health strategies. Keywords Candida spp., MALDI-TOF MS, automated systems, e-health, microbiological diagnostics, bioinformatics. A26 Impact of implementing a standardized documentation system on reporting surgical complications in a surgical oncology department A. Houmada 1 , K. Arroubat 2 , Y. El Bouazizi 3 , A. Souadka 4 , A. Benkabbou 4 , R. Mohsine 5 , M. A. Majbar 4 1 Mohamed V University – National Institute of Oncology, Department of Digestive Oncologic Surgery, Rabat, Morocco; 2 Mohamed V University – National Institute of Oncology, Department of Digestive Oncologic Surgery, Rabat, Morocco; 3 Mohamed V University – National Institute of Oncology, Department of Digestive Oncologic Surgery, Rabat, Morocco; 4 Mohamed V University – National Institute of Oncology, Department of Digestive Oncologic Surgery, Rabat, Morocco; 5 Mohamed V University – National Institute of Oncology, Department of Digestive Oncologic Surgery, Rabat, Morocco BMC Proceedings 2026 , 20(11): A26 Link to published paper 10.1016/j.ejso.2023.107364 A27 Effectiveness of mobile health interventions for symptom management in cancer patients: a systematic review C. Elattabi Mohammed VI University of Health Sciences (UM6SS), Casablanca, Morocco Correspondence: C. Elattabi ([email protected]) BMC Proceedings 2026 , 20(11): A27 Abstract Objective This systematic review aims to evaluate and compare the effectiveness of different mHealth interventions for symptom management in cancer patients, focusing on fatigue, pain, nausea, and anxiety. Methods A systematic search was conducted in PubMed, Embase, Web of Science, and Cochrane Library for studies published up to 2025. Studies involving cancer patients using mobile applications, wearable devices, or telemonitoring interventions were included. Primary outcomes include symptom severity and quality of life, while secondary outcomes include adherence, patient satisfaction, and long-term effectiveness. Results The review will synthesize evidence on the comparative effectiveness of application-based, wearable-based, and telemonitoring interventions. It will also highlight gaps in existing research and explore factors influencing patient engagement and intervention success. Conclusions This systematic review will provide comprehensive evidence on the effectiveness of mHealth interventions in cancer symptom management. The findings aim to guide clinicians, researchers, and developers in designing optimized digital health strategies for cancer care, and to inform future research addressing current gaps in comparative effectiveness and long-term outcomes. Keywords mHealth; cancer; fatigue; symptom management; telemonitoring; wearable devices; mobile applications; systematic review O9 Empowering rural caregivers: a triple-win approach to equity, participation, and sustainability M. Ernst 1 , J. Pflegerl 1 , D. Maurer 2 , A. Schmidt 3 , C. Lampl 3 , J. Goldgruber 4 , S. Dohr 4 , G. Paulinger 5 , P. Plunger 3 , V. Gallistl-Kassing 5 , W. Kratky 4 , E. Turk 1 1 University of Applied Sciences St. Pölten, Austria; 2 Verein Kleinregion Waldviertler Kernland, Ottenschlag, Austria; 3 Gesundheit Österreich GmbH (GÖG), Vienna, Austria; 4 Geriatrische Gesundheitszentren der Stadt Graz (GGZ), Graz, Austria; 5 Karl Landsteiner University of Health Sciences, Krems an der Donau, Austria BMC Proceedings 2026 , 20(11): O9 Background Family caregivers, as backbone of many long-term care systems in Europe, often experience fragmented support networks, limited-service access, and disproportionate mental load. Additionally, multiple crises compound these challenges for family caregivers, including the economic crisis, as well as the climate crisis. Also, societal changes, such as digitalization and socio-ecological transformation often risk leaving vulnerable groups, including family caregivers, behind. Materials and Methods Within the 3WINpA project, promoting climate-conscious practices, social participation, and health promotion, we conducted a participatory Design Thinking process with caregivers and professionals in the Waldviertler Kernland region of Lower Austria based on previously conducted literature review, quantitative survey and qualitative workshops. The goal was to co-create a digital solution addressing inequities in access and coordination. The outcome was the BetreuungsKompass, a hybrid network-mapping and task-coordination prototype designed to make support networks visible, reduce caregiver burden, and overcome barriers related to connectivity, literacy, and access. The co-design process involved three workshops with end-users. First, informal caregivers and health care professionals, including community nurses and regional actors, explored everyday challenges using prototypical personas and developed three concepts to choose from. Secondly, these ideas were assessed for feasibility, potential outcomes, and equity impact. Thirdly, a clickable prototype with administrator and network-partner perspectives was presented and tested. Caregivers and professionals provided structured feedback through click-through sessions and annotated printouts, which informed iterative refinements. Results The workshops demonstrated that visualizing caregiving networks reveals hidden resources and structural gaps, creating opportunities for a fairer distribution of tasks. Caregivers emphasized that coordination often fails not due to unwillingness but because of infrastructural barriers such as unstable internet, shared devices, and limited digital literacy. Professionals highlighted their role in onboarding: community nurses and similar actors from the local care environment can guide caregivers in mapping networks and assist with adding partners. This hybrid approach strengthens trust and accessibility while reducing the mental load borne by primary caregivers. The resulting prototype integrates equity guardrails, including offline functionality and simplified navigation, ensuring that digitally inexperienced users are not excluded. Conclusions This participatory Design Thinking process shows how digital health solutions can be tailored to reduce health disparities in underserved rural contexts. Embedding equity features from the outset and integrating professional actors into onboarding and facilitation, the BetreuungsKompass addresses infrastructural barriers, supports fairer care distribution, and reduces caregiver burden. The digital tool, co-developed with Waldviertler Kernland, is ready for real-world implementation and evaluation of its long-term impact. P1 Management of drug-related risks in pregnant women in community pharmacy: contribution of a digital assistance platform A. Tchimou, S. Mouni, W. Enneffah, A. Hinda, J. Lamsaouri, M. El Wartiti Faculté de Médecine et de Pharmacie, Université Mohammed V, Rabat, 8007, Morocco Correspondence: A. Tchimou ([email protected]) BMC Proceedings 2026 , 20(11): P1 Background Pregnancy is a unique period characterized by profound physiological and psychological changes. The use of medications during pregnancy is frequent but raises considerable anxiety among both pregnant women and healthcare professionals due to the potential teratogenic, embryotoxic, or fetotoxic risks. Community pharmacists, as highly accessible healthcare professionals, are often the first point of contact for pregnant women seeking information or reassurance. Nevertheless, their role is hampered by insufficient continuing education and the lack of practical decision-support tools adapted to the Moroccan context. The aim of this study was twofold: first, to assess the knowledge, attitudes, and practices of pharmacists and pharmacy staff in managing medication use during pregnancy; and second, to develop an interactive website designed as a decision-support platform to enhance community pharmacy practice. Materials and methods A cross-sectional descriptive survey was conducted over a period of 5 months, involving 120 pharmacists, 80 pharmacy assistants, and 150 pregnant women. Two structured questionnaires were used to explore: the training and professional practices of pharmacists and their staff, the difficulties encountered in dispensing medications during pregnancy. Collected data were analyzed descriptively. Based on the findings, a user-friendly website was designed to provide: an overview of the most frequent pregnancy-related conditions, evidence-based hygiene and dietetic recommendations, validated therapeutic drug options considered safe for use during pregnancy, a rapid search engine with an intuitive interface suitable for use at the pharmacy counter. Results The most common conditions reported by pregnant women were nausea and vomiting (90%), constipation (65%), urinary tract infections (53%), and lower back pain (52%). Although 84% of pharmacists reported using databases such as CRAT (Reference Center on Teratogenic Agents), most acknowledged their limited applicability to the Moroccan setting. Furthermore, 72% expressed a strong need for a clear, French-language, locally adapted tool. The developed website was tested by a sample of professionals, of whom 89% rated it as relevant and useful, particularly for its accessibility, time-saving features, and its contribution to safer dispensing and improved counseling. Conclusions Dispensing medications to pregnant women remains a complex and sensitive task requiring vigilance, up-to-date knowledge, and effective communication. The digital tool developed in this study represents an innovative solution that strengthens the pharmacist’s role, provides validated and context-appropriate information, and fosters trust between patients and healthcare providers. Its implementation in community pharmacies could significantly improve drug safety and maternal care in Morocco. P2 The acceptability, benefits and challenges of the new hospital information system “SIH” in Morocco L. Hassani Guennouni 1 , Y. Hamdaoui 2 1 National School of Commerce and Management of Fez, Morocco; 2 Faculty of Legal, Economic and Social Sciences of Fez, Morocco Correspondence: L. Hassani Guennouni ([email protected]) BMC Proceedings 2026 , 20(11): P2 Background As part of the national digitalization of public services, Morocco has introduced a new Hospital Information System (HIS). The success of such initiatives depends largely on staff acceptance, which determines effective implementation and potential impacts on healthcare delivery. Understanding the acceptability, benefits, and challenges of the HIS is therefore essential for guiding future policy and practice. Materials and Methods A cross-sectional survey was conducted among staff in four hospitals to assess perceptions of HIS adoption. Data were collected through a structured questionnaire and analyzed using SPSS. The study was guided by the RE-AIM framework, with a specific focus on the “adoption” dimension. Constructs from the Technology Acceptance Model (TAM), combined with Ajzen’s Theory of Planned Behavior, were used to evaluate perceived usefulness, ease of use, subjective norms, and behavioral control in relation to HIS adoption. Results The majority of respondents were young professionals (68%). Perceived ease of use was significantly associated with age (P = 0.034). Regarding perceived usefulness, 61.5% reported that the HIS facilitated their daily tasks. The overall level of acceptability was approximately 65%, although 23% of participants indicated that system outputs did not adequately meet their information needs. A significant association was also observed between subjective norms and intention to use the system (P = 0.019). Motivation and incentives were highlighted by most respondents as important factors for adoption. Behavioral control was found to positively influence adoption behavior, although it did not show a significant association with intention to use. Conclusions The study highlights moderate levels of HIS acceptability among Moroccan hospital staff, with perceived usefulness and subjective norms playing key roles in adoption. Effective communication of system benefits and the provision of targeted support are essential to address barriers and enhance implementation. These findings suggest that future strategies should focus on strengthening user motivation, improving system outputs, and fostering supportive organizational environments to maximize the potential of HIS in improving public health outcomes. P3 First documentation of nonsyndromic hearing loss due to GJB2 compound heterozygous variants in an Ivorian family M. Toure 1,2 , G. Amalou 1 , I. Ait Raise 1 , N. Max Ange Mobio 3 , A. Malki 2 , A. Barakat 1 1 Genomics and Human Genetics Laboratory, Institut Pasteur du Maroc, Casablanca, Morocco; 2 Ben M’Sik Faculty of Science, Hassan II University of Casablanca, Casablanca, Morocco; 3 ENT Department, University Hospital Medical Center of Treichville, Abidjan, Côte d’Ivoire, Ivory Coast BMC Proceedings 2026 , 20(11): P3 Background The primary etiology of congenital hearing loss is attributed to genetic factors with GJB2 identified as a pivotal gene across diverse ethnic groups. Additionally, nonsyndromic hearing loss is predominantly inherited in an autosomal recessive manner. Materials and Methods We used Sanger sequencing to analyze GJB2 in 17 deaf children from 13 unrelated families in Ivory Coast. Results One family had two children born with severe congenital deafness and exhibited pathogenic compound heterozygous variants. These variants included a nonsense substitution (c.132G>A; p.Trp44Ter) and a newly discovered duplication of seven base pairs (c.205_211dupTTCCCCA; p.Ser72ProfsTer32). Conclusion This study provides the first evidence of GJB2 pathogenic variants causing congenital hearing loss in an Ivorian family. Segregation analysis confirmed the variants and notably, the identified duplication represents a novel mutation that has never been reported worldwide. P4 Study protocol of AI-driven prediction of pediatric diarrheal pathogen dynamics using clinical and metagenomic data in Morocco K. Imani 1,2 , M. Snoussi 1,3 , M. Kettani-Halabi 1,4 , A. Chakib 1,3,5 , I. Diawara 1,2,6 , N. Dini 1,3,5 1 Mohammed VI University of Sciences and Health (UM6SS), Mohammed VI Faculty of Medicine, Laboratory of Microbiology, Infectious Diseases, Allergology and Pathogen Surveillance (LARMIAS), Casablanca, Morocco; 2 Mohammed VI University of Sciences and Health (UM6SS), Mohammed VI Higher Institute of Biosciences and Biotechnologies, Morocco; 3 Cheikh Khalifa International University Hospital, Casablanca, Morocco; 4 Mohammed VI University of Sciences and Health (UM6SS), Mohammed VI Faculty of Pharmacy, Morocco; 5 Mohammed VI International University Hospital, Bouskoura, Casablanca, Morocco; 6 Mohammed VI Center for Research and Innovation, Rabat, Morocco Correspondence: K. Imani ([email protected]) BMC Proceedings 2026 , 20(11): P4 Background Rotavirus continues to be a leading cause of acute diarrhea in children under five despite the introduction of vaccination. In Morocco, current surveillance systems rely mainly on routine laboratory confirmation of cases and do not capture the full microbial etiology or enable predictive monitoring of vaccine impact. The integration of clinical and metagenomic data through artificial intelligence (AI) offers an innovative approach to improve predictive surveillance and public health decision making. Material and Methods We propose a prospective collection of fecal and clinical data from children with acute diarrhea across sentinel sites in Morocco. Clinical variables (e.g., age, vaccination status, dehydration severity) will be recorded in standardized forms. Rotavirus detection, genotyping, and metagenomic sequencing will be conducted, and sequencing outputs will be processed through bioinformatics pipelines to generate structured microbial profiles. Clinical and metagenomic datasets will be merged into a unified database. Machine learning algorithms will be trained to identify predictive associations between clinical presentation, viral genotypes, and co-infections, and to forecast temporal geographic trends of rotavirus circulation. Results This study is expected to generate the first Moroccan dataset combining clinical descriptors with metagenomic surveillance of rotavirus. AI models will be evaluated for their predictive accuracy in identifying high-risk cases and anticipating outbreaks. The integration of vaccination status, viral genotype distribution, and patient severity scores is hypothesized to provide superior predictive capacity compared to traditional surveillance methods. Conclusion The integration of AI with clinical and metagenomic data represents a transformative approach to rotavirus vaccine impact surveillance. This predictive framework will enhance our ability to monitor complex pathogen dynamics, anticipate epidemiological shifts, and assess long-term vaccine effectiveness more effectively than conventional methods. The insights gained will be crucial for informing evidence-based public health interventions, optimizing vaccine policies, and ultimately contributing to improved child health outcomes worldwide, moving from a reactive to proactive approach for managing infectious disease. Keywords Rotavirus, Clinical Data, Metagenomics, Artificial Intelligence, Digital Health, Predictive Surveillance. P5 Transformative technologies for health in Africa: Pathways for biomedical research valorization and technology transfer H. Houssam 1,2 , A. Er-Regragui 1,2 , H. Delsa 3,5,6 , K. Coulibaly 3,5,6 , S. Iskandar 4 , I. Bara 2,8 , A. Bouzyane 1,5,6 , N. Dini 1,5,6 , A. Kettani 9 , I. Diawara 1,2,7 1 Mohammed VI University of Health Sciences (UM6SS), Faculty of Medicine, Laboratory of Microbiology, Infectious Diseases, Allergology and Pathogen Surveillance (LARMIAS), Casablanca, Morocco; 2 Mohammed VI Center for Research and Innovation, Rabat, Morocco; 3 Mohammed VI University of Health Sciences (UM6SS), Faculty of Medicine, Casablanca, Morocco; 4 UM6SS, Faculty of Pharmacy, Casablanca, Morocco; 5 Cheikh Khalifa International University Hospital, Casablanca, Morocco; 6 Mohammed VI International University Hospital, Bouskoura, Casablanca, Morocco; 7 UM6SS, Higher Institute of Biosciences and Biotechnologies, Casablanca, Morocco; 8 UM6SS, Morocco; 9 Faculty of Sciences Ben M’sik, Hassan II University, Casablanca, Morocco BMC Proceedings 2026 , 20(11): P5 Abstract Biomedical research in Africa is increasingly recognized as a catalyst for innovation and public health transformation. However, the effective translation of research outcomes into tangible health solutions remains limited due to fragmented regulatory environments, weak intellectual property (IP) protection systems, and insufficient alignment between research institutions and industry. The emergence of transformative technologies—including digital health platforms, connected medical devices, and biotechnologies—offers new avenues to close this gap. These tools not only support health sovereignty goals but also enable more efficient valorization of scientific knowledge and structured mechanisms for technology transfer. This study draws on a comparative review of scientific publications (2015–2025), policy frameworks, and institutional reports from key stakeholders such as WHO, the African CDC, WIPO, and the African Medicines Agency. A thematic analysis was conducted across three strategic technological pillars: (i) digital health and e-health systems, (ii) Internet of Medical Things (IoMT) and connected devices, and (iii) biotechnologies—including vaccine development and biodiversity valorization. The analysis focused on identifying both systemic barriers and enabling factors for successful research valorization and industrial scale-up. Findings demonstrate that digital health platforms are improving healthcare accessibility in underserved areas and are fostering new ecosystems for health data generation and secondary use. However, these initiatives continue to face challenges related to interoperability, governance, and standardization. The deployment of IoMT and connected devices has shown high promise in areas such as chronic disease management, remote monitoring, and sports medicine, but progress is constrained by limited local manufacturing capacity and underdeveloped data protection frameworks. Biotechnological advances, particularly in vaccine production and bio-therapeutics, illustrate Africa’s potential to align research capacity with industrial and public health needs. Examples from Morocco, South Africa, and Rwanda highlight the role of public–private partnerships in enabling sustainable innovation pathways. Across all three domains, recurring obstacles include insufficient financial investment, limited IP awareness among researchers, and a lack of harmonized regional frameworks to support cross-border innovation transfer. Nevertheless, the establishment of the African Medicines Agency, continental vaccine manufacturing strategies, and growing university–industry partnerships indicate a favorable shift toward systemic support for biomedical innovation in Africa. Transformative technologies offer concrete pathways to accelerate the valorization of African biomedical research and operationalize effective technology transfer. Achieving this requires coordinated strategies that integrate regulatory reforms, regional harmonization, targeted investment, and capacity building across the research-to-industry continuum. By leveraging digital health tools, IoMT infrastructures, and biotechnology platforms, African institutions can reposition themselves as key actors in global health innovation, contributing not only to local healthcare delivery but also to broader international biomedical advancements. Keywords Biomedical research valorization; technology transfer; digital health; IoMT; biotechnology; Africa; innovation ecosystems; health sovereignty P6 Enhancing well-being within the nursing workforce through online interventions: a systematic review S. Bouabid 1,2 , S. Bouftane 5 , A. Chati 3,5 , K. Hassouni 1,2 , S. Belabbes 1,2 , M. Khalis 1,3,4 1 Mohammed VI International School of Public Health, Mohammed VI University of Health Sciences (UM6SS), Casablanca, Morocco; 2 Department of Public Health and Health Management, Mohammed VI Center for Research and Innovation, Rabat, Morocco; 3 Department of Public Health and Clinical Research, Mohammed VI Center for Research and Innovation, Rabat, Morocco; 4 Higher Institute of Nursing Professions and Health Techniques, Ministry of Health and Social Protection, Rabat, Morocco; 5 Faculty of Legal, Economic and Social Sciences of Aïn Chock, Hassan II University of Casablanca, Morocco Correspondence: S. Bouabid BMC Proceedings 2026 , 20(11): P6 Abstract Nurses face increasing levels of occupational stress and burnout, which negatively impact their mental health and the quality of care they provide. In this context, online interventions have emerged as a promising and accessible approach to supporting psychological well-being within the nursing workforce. This systematic review aims to assess the effectiveness of such interventions and identify the key factors influencing their success, thereby informing the development of optimized support strategies for nurses. A systematic literature review was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Searches were carried out in PubMed, Scopus, Web of Science, CINAHL, and PsycINFO, focusing on peer-reviewed studies published between January 2020 and 2024. Included studies targeted nursing professionals and evaluated the impact of online interventions on mental health and well-being outcomes. Thirteen studies met the inclusion criteria. The analysis revealed positive outcomes associated with various forms of digital interventions, including online stress management programs, cognitive-behavioral therapy (CBT) sessions, virtual peer support groups, and mobile applications for well-being monitoring. These interventions were associated with a significant reduction in stress and burnout symptoms, as well as improvements in psychological resilience and job satisfaction. The findings underscore the beneficial role of online interventions in enhancing the well-being of nurses. These digital approaches contribute to reducing psychological distress, fostering emotional resilience, and promoting a healthier work–life balance. Given their accessibility and scalability, such interventions represent an effective component of broader mental health strategies within healthcare systems. Keywords Online interventions; well-being; mental health; nursing staff; burnout; digital health support P7 Repurposing DrugBank compounds as NAD-dependent deacetylase Sirtuin 2 inhibitors via QSAR modelling with gradient boosting algorithms and all-atom molecular simulations Y. Boulaamane 1 , A. Saih 2,3 , A. Guendouzi 4 , A. Maurady 1,5 1 Laboratory of Innovative Technologies, National School of Applied Sciences of Tangier, Abdelmalek Essaadi University, Tetouan, Morocco; 2 Virology Unit, Immunovirology Laboratory, Institut Pasteur du Maroc, Casablanca, Morocco; 3 Laboratory of Biology and Health, URAC 34, Faculty of Sciences Ben M’Sik, Hassan II University of Casablanca, Morocco; 4 Laboratory of Chemistry: Synthesis, Properties and Applications (LCSPA), Faculty of Sciences, University of Saida – Dr Moulay Tahar, Saida, Algeria; 5 Faculty of Sciences and Techniques, Abdelmalek Essaadi University, Tetouan, Morocco Correspondence: Y. Boulaamane ([email protected]) BMC Proceedings 2026 , 20(11): P7 Abstract Sirtuin 2 (SIRT2), a NAD + -dependent histone deacetylase implicated in α-synuclein aggregation, is an emerging target for disease-modifying therapies in Parkinson’s disease (PD). Here, we employed an integrated computational drug-repurposing strategy to identify potent SIRT2 inhibitors from the DrugBank database. A curated set of 949 inhibitors was used to construct quantitative structure–activity relationship (QSAR) models with four gradient-boosting algorithms, yielding CatBoost as the optimal predictor (R²val = 0.74, Q²10-fold = 0.72). The model screened 4,947 drug-like compounds, from which 97 candidates with predicted pIC 50 ≥ 6 were prioritized. Molecular docking against the SIRT2 crystal structure (PDB: 4RMG) revealed high-affinity binding modes for multiple hits, notably DB14822, DB03571, and DB06506, engaging conserved residues (Phe119, Tyr139, Phe190, Ile232) through hydrophobic and π-stacking interactions. ADMET profiling indicated favorable drug-likeness and acceptable pharmacokinetic/toxicity properties for most candidates. All-atom molecular dynamics simulations (250 ns) demonstrated that top ligands maintained compact, stable complexes with low RMSD, restricted radius of gyration, and minimal solvent exposure. Principal component and free-energy landscape analyses confirmed constrained global motions, while MM/GBSA calculations yielded favorable binding free energies (−32.6 to −35.7 kcal/mol) for lead compounds. These results nominate repurposed investigational and approved drugs as promising SIRT2 inhibitors, meriting experimental validation for PD therapy development. Keywords SIRT2 inhibition; Parkinson’s disease; drug repurposing; QSAR; molecular docking; molecular dynamics P8 Biological approaches in evaluating the health risks of ultraviolet radiation, diesel engine exhaust emissions, and other environmental stressors: transforming care through enhanced diagnostics R. Batool, N. Bakht, S. Khan Jogezai, M. W. Khan Department of Biotechnology, Balochistan University of Information Technology, Engineering & Management Sciences (BUITEMS), Quetta, Pakistan BMC Proceedings 2026 , 20(11): P8 Abstract Background Ultraviolet (UV) radiation, diesel engine exhaust (DEE) emissions, and other ambient environmental stressors are major occupational hazards with profound health implications and are classified as known carcinogens by the International Agency for Research on Cancer (IARC). These exposures are associated with immune suppression, DNA damage, and the development of various cancers. Cancer remains a leading cause of mortality worldwide, accounting for nearly ten million deaths annually. In Pakistan, occupational exposure to environmental stressors is intensified by rapid urbanization, extensive diesel vehicle use, limited awareness, socioeconomic constraints, and weak regulatory enforcement. While individual effects of these stressors have been studied, investigations into their combined biological impact remain limited. To the best of our knowledge, this is the first study from the region to examine their combined carcinogenic potential. Materials and Methods This study will include 500 healthy participants, equally divided into exposed and non-exposed groups. The exposed group will consist of outdoor workers occupationally exposed to solar UV radiation and DEE emissions, matched with controls based on gender, age (±5 years), and socioeconomic status. Ambient air pollutants, including PM 10 , PM 2 . 5 , PM 0 . 1 , formaldehyde, and total volatile organic compounds (TVOCs), will be measured using standard portable monitoring devices. Blood samples (3–5 ml) will be collected from each participant to assess DNA damage and oxidative stress biomarkers, including superoxide dismutase, catalase, and malondialdehyde. Statistical analyses, including logistic regression and independent t-tests, will be applied to determine associations between environmental exposures and biological markers. Results It is anticipated that occupationally exposed individuals will exhibit significantly increased DNA damage, reduced antioxidant enzyme activity, and elevated lipid peroxidation compared with controls. These findings are expected to clarify the carcinogenic and oxidative stress-inducing effects of combined occupational exposures. Conclusions The results of this study will contribute to improved assessment of health risks associated with environmental stressors and support the identification of early diagnostic biomarkers. Increased awareness among occupationally exposed workers may facilitate early cancer detection, improving treatment outcomes. Furthermore, the findings will be valuable for healthcare professionals in diagnostic decision-making and for policymakers in prioritizing occupational health and safety measures. Keywords Ultraviolet radiation; diesel engine exhaust; ambient air pollution; biomarkers; DNA damage; cancer; diagnosis; patient outcomes P9 Detecting subtle MLC errors with the PTW 1600SRS detector array: a GPR-based picket fence approach M. Driouch 1,2 , Y. Adib 3 , A. S. A. Almaamari 4 , M. A. Youssoufi 5 , E. Chakir 1 , E. Al Ibrahmi 1 1 LPMS, Faculty of Sciences, Ibn Tofail University, Kenitra, Morocco; 2 Clinique Avicenne de Fès, Fès, Morocco; 3 LPHE-MS, Faculty of Science, Mohammed V University, Rabat, Morocco; 4 Faculty of Medicine, Mohammed V University, Rabat, Morocco; 5 Department of Radiotherapy, National Institute of Oncology, UHC Ibn Sina, Mohammed V University, Rabat, Morocco BMC Proceedings 2026 , 20(11): P9 Background Modern radiotherapy techniques like VMAT and SBRT rely on precise Multi-Leaf Collimator (MLC) motion for dose delivery, where even small positioning errors can have significant clinical consequences. Traditional MLC quality assurance tests, such as the picket fence, often use film or EPID methods that involve subjective analysis. This study investigates the PTW 1600SRS Detector Array with Gamma Passing Rate (GPR) analysis to establish a more objective and quantitative QA methodology. The array's high spatial resolution makes it particularly suitable for evaluating MLC performance, aiming to improve detection of subtle inaccuracies and enable reliable performance tracking. Materials and methods The study utilized a high-resolution PTW SRS1600 detector array containing 1,521 ionization chambers. Measurements were performed on an Elekta linac equipped with an Agility MLC system (160 leaves, 5 mm width). The picket fence test was designed with 9 dynamic pickets (3 mm and 5 mm apertures) and delivered using 6 MV FF beams. The PTW array was centered at isocenter between PMMA slabs (2 cm above, 5 cm below) at SSD = 100 cm. All plans were calculated in MONACO® TPS with a 2 mm grid and delivered via a VersaHD® linac. Results Analysis of the picket fence test using the PTW SRS1600 array revealed a characteristic decrease in Gamma Passing Rate (GPR) at the central region for the 3 mm gap width, with values ranging from 81.3% (2%/2 mm) to 54% (1%/1 mm). This phenomenon is attributed to the source occlusion effect in small-field dosimetry, where beam convergence causes penumbral overlap and higher central dose delivery. The 2%/2 mm criterion demonstrated superior and more stable GPR results compared to the stricter 1%/1 mm criterion. The high spatial resolution of the PTW SRS1600 array proved essential for detecting these subtle MLC positional variations, outperforming conventional detector arrays and providing an objective method for tracking MLC performance over time. Conclusion The PTW SRS1600 detector array proves to be a highly effective tool for objective and efficient Multi-Leaf Collimator quality assurance. By enabling quantitative Gamma Passing Rate analysis, this method replaces subjective assessments with reproducible data, allowing for precise detection of subtle MLC inaccuracies—such as the central axis deviation in the picket fence test—and reliable tracking of MLC performance over time. This approach significantly enhances the robustness and precision of QA programs for modern radiotherapy. P10 Revolutionary AI systems transform colorectal cancer detection: a new era in medical imaging H. Elmarrachi 1,2 , M. Andrif 1,2 , N. Ismaili 1,2,3 1 Mohammed VI Faculty of Medicine, Mohammed VI University of Health Sciences, Casablanca, Morocco; 2 Research Unit, Mohammed VI Center for Research and Innovation, Rabat, Morocco; 3 Cheikh Khalifa International University Hospital, Mohammed VI University of Health Sciences, Casablanca, Morocco Correspondence: H. Elmarrachi BMC Proceedings 2026 , 20(11): P10 Abstract The detection of colorectal cancer is entering a transformative phase with the rapid advancement and clinical adoption of artificial intelligence (AI) technologies. What was once experimental is now becoming an integral component of medical imaging, with AI systems demonstrating the capacity to improve diagnostic accuracy, standardize assessments, and support clinical decision-making across multiple imaging modalities. This review synthesizes findings from studies published between 2018 and 2024, focusing on AI platforms designed to distinguish malignant colorectal lesions from benign ones. The analysis spans applications in endoscopy, radiology, and digital pathology, with emphasis on diagnostic performance metrics and readiness for clinical implementation. AI-assisted colonoscopy systems now achieve real-time sensitivity exceeding 97%, enabling enhanced detection of polyps and early-stage malignancies during routine procedures. Capsule endoscopy platforms integrated with AI demonstrate precision rates surpassing 95%, optimizing non-invasive diagnostics. In radiology, advanced AI tools for CT imaging provide robust tumor identification and segmentation capabilities, while AI-enabled digital pathology systems consistently report diagnostic accuracy above 95%, supporting faster and more reliable histopathological interpretations. These findings underscore the readiness of AI systems for clinical integration and their potential to redefine colorectal cancer screening and diagnostic workflows. Among current applications, AI-powered endoscopic tools are the most advanced in real-world deployment, while radiology and pathology solutions are rapidly advancing toward routine clinical use. AI-enhanced colorectal cancer detection represents a paradigm shift, moving from research settings to transformative clinical tools. Their widespread adoption offers the potential to significantly improve early detection, reduce diagnostic variability, optimize treatment planning, and ultimately improve patient outcomes. Keywords Artificial intelligence; colorectal cancer; medical imaging; deep learning; endoscopy; digital pathology; cancer detection; clinical applications P11 Immersive simulation in nursing education: effects on engagement, motivation, satisfaction, and self-confidence L. Ben Yahya 1 , M. Radid 1 , M. El Yaagoubi 2 , L. El Moumou 3 , O. Abouri 4 , A. Naciri 1 , G. Chemsi 5 1 Laboratory of Sciences and Technologies of Information and Education, Faculty of Sciences Ben M’Sik, Hassan II University of Casablanca, Casablanca, Morocco; 2 High Institute of Nursing Professions and Health Techniques, ISPITS Agadir – Annex Tiznit, Morocco; 3 Biotechnology, Environment and Health Team, Laboratory of Sciences of Health and Environment, ISPITS Agadir – Annex Tiznit, Morocco; 4 Laboratory of Inflammatory Cellular and Molecular Physiopathology, Degenerative and Oncological, Faculty of Medicine and Pharmacy, Hassan II University of Casablanca, Casablanca, Morocco; 5 Laboratory of Mathematics, Artificial Intelligence, and Digital Learning, Hassan II University of Casablanca, Casablanca, Morocco BMC Proceedings 2026 , 20(11): P11 Abstract Virtual reality (VR) technologies are playing an increasingly important role in nursing education by enhancing experiential learning, supporting clinical decision-making, and improving the application of theoretical knowledge in practice. Immersive simulation-based learning (ISBL) is one such VR-enhanced approach, designed to boost learner engagement, motivation, satisfaction, and self-confidence. This study explores the impact of ISBL on nursing students, with a specific focus on its effectiveness in anatomy education. A quasi-experimental study was conducted from January to February 2025 with 76 nursing students, who were randomly assigned to either an experimental group receiving immersive simulation (n = 38) or a control group following traditional instructional methods (n = 38). A pre- and post-intervention test design was used to assess changes in student motivation, engagement, satisfaction, and self-confidence. Statistical analyses were conducted using non-parametric tests, specifically the Mann–Whitney U test and Wilcoxon signed-rank test, via IBM SPSS. The results revealed that students in the immersive simulation group showed significantly greater improvements in motivation (Z = -4.407, p < 0.001), engagement (Z = -3.555, p < 0.001), and self-confidence (Z = -2.054, p = 0.040) compared to the traditional instruction group. However, the difference in learning satisfaction between the two groups did not reach statistical significance (Z = -1.660, p = 0.097). These findings suggest that immersive simulation contributes meaningfully to enhancing nursing students’ motivation, engagement, and self-confidence. While overall satisfaction levels remained similar across both groups, the use of immersive simulation emerges as a valuable supplement to traditional teaching methods and holds promise for addressing pedagogical challenges in healthcare education, particularly in resource-constrained contexts such as Morocco. Keywords Simulation-based learning; immersive simulation; nursing education; virtual reality; Morocco P12 Detection of hyperandrogenism anovulation using artificial intelligence: deep learning approach and model explainability A. Elhachimia 1 , A. Benksimb 2 , M. Eddabbah 3 , M. Cherkaouia 1 1 Département de Biologie, Université Cadi Ayyad de Marrakech (UCAM), Marrakech, Morocco; 2 Institut des Professions Infirmières et des Techniques de Santé (ISPITS), Marrakech, Morocco; 3 École Supérieure de Technologie d’Essaouira (ESTE), Université Cadi Ayyad, Essaouira, Morocco Correspondence: M. Cherkaouia BMC Proceedings 2026 , 20(11): P12 Background Hyperandrogenic anovulation (HA) is a common endocrine disorder and a major cause of female infertility. Its diagnosis remains challenging due to the clinical heterogeneity of symptoms and the absence of standardized diagnostic criteria. This often leads to underdiagnosis and delays in management. Recent advances in artificial intelligence (AI) offer promising tools for enhancing early detection of such complex conditions by integrating clinical, metabolic, and hormonal data. Materials and Methods A retrospective dataset of 541 patients was used, including 45 features such as age, BMI, FSH, LH, AMH levels, and menstrual cycle regularity. A rigorous data preprocessing pipeline was implemented: missing values were imputed, features were scaled using min-max normalization, and outliers were removed based on z-score and interquartile range (IQR) methods. Four deep learning models were developed and compared: feedforward neural network (FNN), convolutional neural network (CNN), recurrent neural network (RNN), and long short-term memory (LSTM). Model training involved 5-fold cross-validation and hyperparameter optimization. Model explainability was assessed using SHAP (SHapley Additive exPlanations), while class imbalance was addressed using SMOTE (Synthetic Minority Over-sampling Technique). Results Among the tested models, the RNN achieved the highest performance with an accuracy of 95.8% and a recall of 94%, followed closely by CNN (95%) and LSTM (94.6%). The FNN showed lower performance (92.3%). SHAP analysis revealed that the most important predictors were the FSH/LH ratio, AMH concentration, BMI, and follicle count. These variables contributed significantly to the model's ability to distinguish HA cases. The integration of SMOTE improved classification metrics for the minority class without degrading overall performance. Conclusions This study highlights the potential of deep learning models, particularly RNNs, to support early and reliable detection of hyperandrogenic anovulation based on multi-dimensional clinical data. The high accuracy, combined with the use of SHAP for interpretability and SMOTE for fairness, suggests that such AI-based tools could be integrated into clinical decision-making workflows. This is especially valuable in low-resource settings where access to expert endocrinological assessment is limited. Further validation on external cohorts is needed to generalize these findings. P13 Applications of machine learning in palliative care: a narrative review A. AitOuma 1,2 1 Faculty of Medicine and Pharmacy, Mohammed V University, Rabat, Morocco; 2 Ibn Sina University Hospital Center, Rabat, Morocco BMC Proceedings 2026 , 20(11): P13 Abstract Background The rapid aging of the global population, combined with the increasing prevalence of chronic and degenerative diseases such as cancer, organ failure, and dementia, is creating an exponential demand for palliative care. This demographic and epidemiological transition calls for the transformation of clinical practices toward more predictive, personalized, and interdisciplinary approaches. The objective of this review was to analyze recent applications of machine learning in palliative care, identifying opportunities, limitations, and ethical challenges. Methods This narrative review explored applications of machine learning in palliative care published between 2020 and 2025. Scientific articles were retrieved from PubMed, Cochrane, Scopus, SpringerLink, and ScienceDirect using the keywords “Machine Learning,” “Artificial Intelligence,” and “Palliative Care.” After manual screening for clinical relevance, methodological rigor, and direct linkage to palliative care, nine articles were included. Results Machine learning algorithms were applied to predict mortality and end-of-life trajectories, stratify patient profiles, automatically detect symptoms, care goals, or psychological distress, improve analgesic prescription and triage, and analyze patient preferences and ethical issues through clinical text mining. Reported clinical benefits included improved prognostic accuracy and care personalization. However, challenges were identified, including heterogeneous data quality in electronic health records, limited transparency and explainability of algorithms, insufficient multicenter clinical validation, and the difficulty of integrating technological solutions into highly human-centered contexts. Conclusions Machine learning represents a promising technological advance in palliative care, but its integration requires a cautious, ethical, and collaborative approach. Future success will depend on joint efforts by researchers, clinicians, patients, and policymakers to ensure clinical utility while respecting the core values of palliative care. Keywords Palliative care; machine learning; artificial intelligence; prediction; clinical decision support P14 AI at work: preventing occupational risks or creating them? M. Lghabi 1 , B. Benali 2 1 Maître de conférences en médecine du travail, Faculté de médecine et de pharmacie de Marrakech, Université Cadi Ayyad, Morocco; 2 Professeur de médecine du travail, Faculté de médecine et de pharmacie de Rabat, Université Mohamed V, Morocco BMC Proceedings 2026 , 20(11): P14 Background Artificial intelligence (AI) is profoundly reshaping professional environments, offering unprecedented opportunities for risk prevention while simultaneously generating new hazards. The objective of this study is to analyze the dual impact of AI on occupational safety and health (OSH). Materials and Methods The data analyzed derive from the scientific literature as well as from reports and publications issued by organizations and institutions specializing in occupational safety and health, including the ILO, INRS, and EU-OSHA. Results AI provides significant benefits for OSH, including continuous worker monitoring through wearable devices, intelligent building systems, automated hazard detection, smart personal protective equipment, workplace violence monitoring, automated substance screening, mental health monitoring, prevention of musculoskeletal disorders, automation of hazardous tasks, automated compliance audits, and decision-support systems. These technologies facilitate distancing from high-risk environments, reduce physical strain, and improve predictive monitoring. However, AI also generates new risks: mechanical failures, ergonomic issues, work intensification, technostress, social isolation, privacy violations, excessive surveillance, algorithmic discrimination, and job insecurity. Psychosocial risks include loss of autonomy, cognitive overload, and anxiety linked to automation. Conclusion AI in occupational safety and health holds considerable potential for prevention and efficiency, but its deployment must be governed by ethical and participatory frameworks. Only a balanced integration of technological innovation and respect for workers’ rights will allow it to serve as a genuine driver of progress. Keywords Artificial intelligence; occupational safety and health; occupational risks; prevention Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Change history 4/15/2026 A Correction to this paper has been published: 10.1186/s12919-026-00376-2 Articles from BMC Proceedings are provided here courtesy of BMC ACTIONS View on publisher site PDF (942.1 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top