CARE: Confidence-Aware Reasoning for Reliable Medical VQA

CARE: Confidence-Aware Reasoning for Reliable Medical VQA Yuetian Du1† , Yucheng Wang1† , Zhenyuan Chen1 , Luyuan Chen1 , Rongyu Zhang1 , Jinjian Zhang2 , Wei Zhou2 , Zhijie Xu3 , Ming Kong1 , Zhan Zhou1 , Jie Liu4B , and Qiang Zhu1B Zhejiang University 2 Ant Group University of Michigan 4 City University of Hong Kong {22421227,zhuq}@zju.edu.cn 1

arXiv:2608.10964v1 [cs.CV] 11 Aug 2026

3

Abstract. Reinforcement Fine-Tuning (RFT) has enabled medical Multimodal Large Language Models (MLLMs) to produce Chain-of-Thought (CoT) reasoning for visual question answering, yet these models suffer from confidence miscalibration—a systematic gap between expressed certainty and actual diagnostic accuracy that undermines clinical trust. We propose CARE, a Confidence-Aware medical REasoning framework that jointly optimizes accuracy and calibration through a dual-stage pipeline. First, a scalable Medical-CoT synthesis provides structured cold-start data for Supervised Fine-Tuning. Second, Group Relative Policy Optimization (GRPO) with a novel Confidence-Aware Reward (CAR) mechanism ties the model’s confidence to diagnostic correctness within the reward signal. Across three Medical VQA benchmarks, CARE achieves the highest diagnostic accuracy while obtaining the lowest Expected Calibration Error and Hallucination Rate, establishing a foundation for trustworthy clinical decision support. Our code is available at https://github.com/anotherbricki/CARE. Keywords: Reinforcement Fine-Tuning · Medical VQA · Confidence Calibration.

1

Introduction

Medical Multimodal Large Language Models (MLLMs) are increasingly applied to clinical decision support tasks such as medical visual question answering (VQA) [12]. Typically built through Supervised Fine-Tuning (SFT) on curated clinical instruction datasets, these models learn a direct input-output mapping that produces predictions without transparent reasoning processes. In clinical practice, where diagnostic errors carry severe consequences, physicians are unlikely to trust model predictions that lack interpretable reasoning chains. This highlights an urgent need for medical MLLMs that can perform transparent, step-by-step clinical reasoning. †

Equal contribution.

B

Corresponding author.

2

Y. Du et al.

Answer from MedVLM-R1 Unconfident! Proportion(%)

[Begin of CoT] The image shows a lateral view of the lungs ··· this is a typical manifestation of pneumonia. Pneumonia ··· which are consistent with these imaging features. In addition, there are no obvious signs of pleural effusion or pneumothorax, further supporting the diagnosis of pneumonia. Taken together, the patient has pneumonia. Shallow-Thinking [End of CoT] <answer> Pneumonia. </answer>

Confidence

·Wrong Diagnosis

[Begin of CoT] 1. Identify the Anatomical Structures: The image shows a cross-sectional view of the abdomen ··· which appears distended and abnormal in shape and size, suggesting a blockage or obstruction. 2. Assess Intestinal Distension: ··· 6. Conclusion: Based on ··· the patient is most likely suffering from blind loop syndrome ··· Summary: The patient's condition ···strongly suggests blind loop syndrome, characterized by significant distension of the small intestine loops, indicative of an obstruction. [End of CoT] Reasonable-Thinking <answer> Blind loop syndrome. </answer>

Proportion(%)

Answer from CARE Model (ours) Question: Based on the visual information provided, could you tell me what the current condition of the patient is?

Confident!

Confidence

·Confidence-Aware ·Well-Structured ·High-Reliability

Fig. 1. Comparative Analysis of Diagnostic Reasoning Paths. Existing medical reasoning MLLMs such as MedVLM-R1 exhibit confidence miscalibration, where expressed certainty fails to reflect actual diagnostic accuracy. CARE explicitly aligns subjective confidence with diagnostic correctness, producing verifiable and interpretable reasoning trajectories for clinical decision-making.

Reinforcement Fine-Tuning (RFT), where RL algorithms such as Group Relative Policy Optimization (GRPO) [16] optimize models against verifiable reward signals, has recently enabled reasoning-focused models that generate extended Chain-of-Thought (CoT) outputs. This paradigm has been adopted in the medical domain, producing early medical reasoning models [9,15] for Medical VQA. While these models improve reasoning transparency, a critical issue persists: confidence miscalibration [21,20,5]. As illustrated in Figure 1, existing models exhibit a systematic misalignment between expressed confidence and actual diagnostic accuracy: they may appear underconfident on correct diagnoses or assign unwarranted certainty to incorrect ones. In clinical deployment, such miscalibration undermines trust, as a model that cannot reliably reflect its own uncertainty provides no safe basis for decision-making. Related Work. Reinforcement Fine-Tuning (RFT) has emerged as a new paradigm for improving reasoning in large models [13,22], surpassing traditional SFT-based CoT approaches. Unlike SFT, which passively fits existing data distributions, RFT enables models to receive direct feedback from verifiable outcomes and dynamically refine their reasoning paths [2]. Pioneered in mathematical reasoning [16] and popularized by DeepSeek-R1 [3], RFT has recently been extended to multimodal medical applications, producing models such as Med-R1 [9], MedVLM-R1 [15], and others [18,4,19,8] for clinical VQA. However, existing RFT frameworks optimize primarily for answer correctness through verifiable rewards [10], without explicitly addressing the alignment between model confidence and prediction accuracy. Unlike methods that elicit verbal confidence or simply reward high confidence, CARE uses confidence as a correctness-conditioned calibration signal within GRPO. Medical-CoT synthesis mainly provides a verified cold start that stabilizes reasoning format and answer extraction for subsequent confidence-aware RL. To address this gap, we propose CARE—a Confidence-Aware medical REasoning framework that jointly optimizes diagnostic accuracy and confidence

CARE: Confidence-Aware Reasoning for Reliable Medical VQA

3

calibration within a unified two-stage RFT pipeline. Our main contributions are as follows: • Confidence-Aware Reinforcement Fine-Tuning: We integrate a novel Confidence-Aware Reward (CAR) into the GRPO framework, explicitly aligning the model’s expressed confidence with diagnostic accuracy during RL optimization. • Scalable Medical-CoT Data Construction: We design an automated synthesis pipeline to construct high-quality medical reasoning data for multiple clinical scenarios, providing structured diagnostic trajectories with verifiable conclusions. • Consistent Improvements in Accuracy and Calibration: Evaluations across multiple Medical VQA benchmarks demonstrate that CARE achieves superior diagnostic accuracy while substantially reducing confidence miscalibration, establishing a reliable foundation for clinical decision support.

2

Method

2.1

Overview

As illustrated in Figure 2, the CARE framework consists of two phases: (1) Medical-CoT Data Synthesis, which constructs scalable, well-structured diagnostic reasoning paths from existing clinical datasets; and (2) Two-Stage Optimization, transitioning from SFT-based domain adaptation to GRPO reinforcement learning driven by a novel Confidence-Aware Reward (CAR) mechanism. This design jointly optimizes diagnostic accuracy and confidence calibration, mitigating the risk of confidence miscalibration in clinical judgments. 2.2

Scalable Medical-CoT Data Synthesis

To facilitate structured reasoning without relying solely on prohibitive expert annotations, we design an automated, reverse-thinking synthesis pipeline. Let DVQA = {(Vi , Qi , Yi )}N i=1 denote a standard Medical-VQA dataset comprising visual contexts V , text queries Q, and ground truth diagnoses Y . We employ a capable base MLLM, denoted as πθ , to retroactively generate intermediate reasoning trajectories T : Ti ∼ πθ (Vi , Qi , Yi ). (1) To ensure the clinical validity and logical coherence of Ti , we enforce a strict structural template requiring explicit phase identifiers (e.g., visual analysis, differential diagnosis) and a conclusive summary. Crucially, rather than relying on raw generative outputs, we implement a rigorous verification mechanism. An auxiliary verifier (e.g., GPT-4o) evaluates each trajectory against the ground truth Yi . A generated trajectory is admitted into the final training corpus DCoT if and only if its reasoning chain logically deduces a conclusion that perfectly aligns with Yi . This objective filtering acts as a scalable proxy for quality assurance, ensuring the model learns from goal-oriented, logically sound diagnostic paths.

4

Y. Du et al. Medical-CoT Data Construction

Two-Stage CARE Training Framework A1

We carefully selected 4 types of visual questions for this dataset.

A2

VLMs

GRPO

…

…

X-ray CT MRI Pathology

…

Question + Images

Image types • • • •

STAGE 2: Confidence-Aware RFT

STAGE 1: CoT-Driven SFT

Sample Selection

An

Reasoning Process Final Answer

MLLM Generation

Reward Evaluation Accuracy/Recall

Format

CAR

Question: What is the condition of the patient?

Answer: Anatomical Analysis: The image shows a cross-sectional view of the abdomen, likely from a CT scan ... Summary of Findings: The patient's condition ... strongly suggests blind loop syndrome ... Final Answer: Blind Loop Syndrome

Quality Curation Reasoning

Consistency or not? MLLM Verifier

Ground Truth

Confidence-Aware Reward (CAR) Reasoning Process

avg prob Confidence

This

chest

inspir

atory

X

-ray effort

indicates

adequate

Final Answer: Yes, this does represent adequate inspiratory effort.

Well-Calibrated!

�����������n

… , verify

Accuracy

������������

= ���� × ���� − (� − ���� ) × ����

Fig. 2. Overview of the CARE Framework. The pipeline consists of two phases: (1) Medical-CoT Data Construction for synthesizing structured diagnostic reasoning paths, and (2) Two-Stage Optimization, comprising SFT cold-start and GRPO with the proposed Confidence-Aware Reward (CAR) mechanism to align confidence with diagnostic accuracy.

2.3

Two-Stage Optimization of CARE

Leveraging the curated DCoT , CARE is trained in two stages: SFT cold-start for domain adaptation, followed by confidence-aware GRPO-based reinforcement learning. Phase I: SFT Cold Start. We conduct standard supervised fine-tuning on DCoT to establish the model’s capacity for structured clinical reasoning. For each tuple (V, Q, T , Y ) ∈ DCoT , the model maximizes the likelihood of the target sequence S = [T ; Y ]:   |S| X LSFT (θ) = −E(V,Q,S)∼DCoT  log πθ (st |V, Q, s<t ) .

(2)

t=1

This phase yields a reference policy πref that follows the structured reasoning format, serving as initialization for the RL phase. Phase II: GRPO-based RL. We then optimize πθ using GRPO, which eliminates the need for a separate value network. For each query (V, Q), the policy samples K candidate outputs {o1 , . . . , oK }. The objective maximizes grouprelative advantages while constraining divergence from πref : "

# K 1 X JGRPO (θ) = E min (ρi , clip(ρi , 1 − ϵ, 1 + ϵ)) Ai − βDKL (πθ ∥πref ) , (3) K i=1

CARE: Confidence-Aware Reasoning for Reliable Medical VQA

5

θ (oi |V,Q) where ρi = ππold is the importance weight, ϵ is the clip ratio, and β controls θ (oi |V,Q) the KL penalty. The advantage Ai is computed by normalizing the composite reward Ri within the sampled group:

Ai =

2.4

Ri − µ(R1:K ) . σ(R1:K )

(4)

Confidence-Aware Reward (CAR) Formulation

Existing RFT frameworks rely on binary format or accuracy rewards, leaving confidence miscalibration unaddressed. We introduce the Confidence-Aware Reward (CAR) to close this gap by directly incorporating confidence alignment into the reward signal. The composite reward Ri = Rform + Rout + Rcalib consists of three components: Format & Output Rewards (Rform , Rout ). Rform ∈ {0, 1} enforces the use of designated <think> and <answer> delimiters. Rout evaluates diagnostic correctness: for closed-ended tasks, it is an exact-match indicator I(Y ⊆ ai ); for open-ended tasks, it uses a recall-based metric to assess the coverage of ground truth Y within the predicted answer ai . Calibration Reward (Rcalib ). This is the core component that distinguishes CARE from standard RFT. We use the model’s answer-level predictive confidence as a calibration signal, rather than treating token probability as a complete epistemic uncertainty estimate. Specifically, for each output oi , let ai = {t1 , . . . , t|ai | } denote only the tokens inside the <answer> span, excluding reasoning tokens, formatting tokens, and special tokens. We compute: |ai | 1 X πθ (tj | V, Q, t<j ). C(ai ) = |ai | j=1

(5)

The calibration reward ties this answer-level confidence to diagnostic correctness: Rcalib (oi , Y ) = Rout · C(ai ) − λ(1 − Rout ) · C(ai ),

(6)

where λ controls the penalty on overconfident incorrect predictions. When the prediction is correct, higher answer confidence is rewarded; when it is incorrect, high confidence is penalized. This correctness-conditioned design differs from simply encouraging high confidence, and directly optimizes the confidence-accuracy alignment measured by ECE.

3

Experiments

3.1

Experimental Setup

Datasets. We evaluate CARE on three widely adopted Medical VQA benchmarks:

6

Y. Du et al.

• VQA-RAD [11]: Focused on radiology, comprising 315 clinician-annotated images and 3,515 query-answer pairs. • SLAKE [14]: A semantically-labeled, knowledge-enhanced dataset featuring 642 images and 14,000 bilingual pairs. • PathVQA [7]: Dedicated to pathology, containing 4,998 images and 32,799 question-answer pairs. Baselines. We evaluate CARE against state-of-the-art reasoning-focused Medical MLLMs, categorized by parameter scale: (1) Compact-Scale Models (≤ 3B): Including Med-R1-3B [9] and MedVLM-R1-2B [15]. (2) Standard-Scale Models (7B–8B): Including Lingshu-7B [19], MedVLThinker-7B [8], Fleming-VL-8B [17], and MedMO-8B [4]. Evaluation Metrics. We assess model performance along three dimensions: • Diagnostic Accuracy: We report Accuracy for closed-ended questions and Recall for open-ended questions, following standard Medical-VQA evaluation protocols. • Confidence Calibration: We adopt Expected Calibration Error (ECE) [6] to measure the alignment between model confidence and actual accuracy. Confidence scores are partitioned into M equally spaced bins Bm , and ECE computes the weighted deviation: ECE =

M X |Bm | m=1

N

|acc(Bm ) − conf(Bm )|,

(7)

where N is the total number of samples. Lower ECE indicates better calibration. • Hallucination Rate: We employ Lingshu-32B [19] as a VLM judge to evaluate reasoning trajectories under a fixed prompt and rubric. The judge is run with deterministic decoding, and all methods are evaluated with identical inputs to ensure fairness. Each sample receives a normalized hallucination score ri , and the overall HR is computed as: HR =

C h

2

E (1 − ri ) ti

i,

(8)

where ti is the output token count and C is a global normalization constant ensuring HR ∈ [0, 1]. The squared term (1−ri )2 applies a quadratic penalty that disproportionately suppresses samples with high hallucination rates, amplifying the contribution of heavily hallucinated outputs to the overall score. 3.2

Implementation Details

CARE is built upon Qwen2.5-VL-7B-Instruct [1]. The same model architecture serves as both the CoT data synthesizer in Section 2.2 and the policy model for subsequent training, ensuring consistency between the generated reasoning format and the optimization target. Training is conducted via full-parameter

CARE: Confidence-Aware Reasoning for Reliable Medical VQA

7

fine-tuning on 6× NVIDIA A100 GPUs. During the SFT Cold Start, we use the AdamW optimizer with a learning rate of 1 × 10−5 and cosine annealing. In the GRPO RL Phase, we generate K = 4 rollouts per query with a batch size of 2 and bfloat16 mixed precision. In the calibration reward of CAR, the penalty coefficient is set to λ = 0.5. Table 1. Main results on Medical VQA benchmarks. The best and second-best results in each column are highlighted in bold and underlined, respectively. The same applies to Table 2 (ACC = accuracy, ECE = expected calibration error, HR = hallucination rate). Model

VQA-RAD

SLAKE

PathVQA

ACC ↑ ECE ↓ HR ↓ ACC ↑ ECE ↓ HR ↓ ACC ↑ ECE ↓ HR ↓ Compact-Scale Models (≤ 3B) Med-R1-3B MedVLM-R1-2B

0.513 0.532

0.451 0.195 0.448 0.200

0.596 0.504

0.369 0.257 0.474 0.237

0.387 0.387

0.584 0.155 0.587 0.168

0.816 0.831 0.678 0.819

0.277 0.322 0.424 0.181

0.563 0.619 0.652 0.629

0.383 0.522 0.569 0.455

Standard-Scale Models (7B–8B) MedMO-8B Lingshu-7B MedVLThinker-7B Fleming-VL-8B

0.647 0.679 0.637 0.668

0.264 0.375 0.413 0.333

0.365 0.083 0.071 0.073

0.368 0.105 0.099 0.089

0.223 0.098 0.082 0.073

CARE-7B (Ours) 0.767 0.202 0.048 0.873 0.115 0.070 0.689 0.290 0.059

3.3

Experimental Results and In-depth Analysis

CARE simultaneously achieves the best accuracy, calibration, and lowest hallucination across all benchmarks. As shown in Table 1, existing medical reasoning models consistently exhibit trade-offs among these three dimensions: strong accuracy often comes with poor calibration or high hallucination, and vice versa. No baseline ranks first across all three metrics on any single benchmark. CARE is the only model to do so consistently, achieving the highest accuracy (e.g., 0.873 on SLAKE, 0.767 on VQA-RAD) while simultaneously obtaining the lowest ECE and HR. The substantial ECE reduction (e.g., 0.115 on SLAKE, a 36% relative improvement over the second-best Fleming-VL-8B) directly confirms the effectiveness of CAR in aligning confidence with accuracy. The consistently lowest HR (e.g., 0.048 on VQA-RAD) further validates our reverse-thinking data synthesis: answer-grounded reasoning paths reduce fabrication at the source by providing structured, goal-oriented CoT trajectories. Different question types favor different training configurations. Table 2 reveals a clear pattern: for closed-ended questions, RL alone achieves the best accuracy and ECE across all three benchmarks (e.g., 0.864 ACC and 0.096 ECE on VQA-RAD), as the constrained answer space allows RL exploration

8

Y. Du et al. Table 2. Ablation study results for different training stages. VQA-RAD

Training Stages

SLAKE

PathVQA

Open ↑ ECE ↓ Closed ↑ ECE ↓ Open ↑ ECE ↓ Closed ↑ ECE ↓ Open ↑ ECE ↓ Closed ↑ ECE ↓ Training-Free SFT RL SFT+RL

0.483 0.520 0.563 0.620

0.485 0.468 0.407 0.363

(a) VQA-RAD

0.658 0.629 0.864 0.658

0.341 0.359 0.096 0.323

0.513 0.803 0.819 0.881

0.456 0.188 0.176 0.112

(b) SLAKE

0.690 0.767 0.861 0.779

0.287 0.225 0.119 0.209

0.150 0.349 0.166 0.421

0.823 0.640 0.727 0.561

0.664 0.857 0.955 0.863

0.326 0.133 0.020 0.121

(c) PATH-VQA

Fig. 3. Confidence distributional shift. Red and green bars represent the normalized confidence scores pre- and post-training, respectively. After training, the mean shifts rightward with increased magnitude, reflecting improved confidence-accuracy alignment.

to converge efficiently without SFT initialization. For open-ended questions, however, SFT+RL consistently dominates (e.g., 0.881 on SLAKE Open, 0.421 on PathVQA Open), since generating free-form diagnostic reasoning requires the structured CoT foundation established during SFT. Notably, SFT alone already yields substantial open-ended gains (e.g., SLAKE Open rises from 0.513 to 0.803), confirming that the CoT cold-start is the primary driver for open-ended performance, while RL further refines calibration on top of it. The results also reflect the complementary roles of the two stages: Medical-CoT provides a stable reasoning and answer-extraction format, especially for open-ended questions, while the RL stage further optimizes answer correctness and calibration through CAR. CAR improves confidence-accuracy alignment rather than merely increasing confidence. Figure 3 compares the normalized confidence distributions before and after CAR alignment. The post-training distribution shifts toward higher-confidence regions, but this shift should be interpreted together with the simultaneous ECE reductions in Table 1. Since ECE directly measures the gap between confidence and empirical accuracy, the improved ECE indicates that the confidence shift is better aligned with diagnostic correctness rather than being a simple increase in likelihood. This supports the design of CAR, which rewards confident correct predictions while penalizing overconfident incorrect ones.

4

Conclusion

We present CARE, a framework that addresses confidence miscalibration in RFTbased medical reasoning models. By combining a scalable Medical-CoT synthesis pipeline with a Confidence-Aware Reward (CAR) mechanism within GRPO,

CARE: Confidence-Aware Reasoning for Reliable Medical VQA

9

CARE explicitly aligns the model’s expressed confidence with diagnostic accuracy during RL optimization. Experiments across three Medical VQA benchmarks show that CARE achieves the best diagnostic accuracy while simultaneously obtaining the lowest ECE and Hallucination Rate, demonstrating that accuracy and calibration can be jointly improved rather than traded off. These results suggest that incorporating confidence signals into the reward design is an effective path toward reliable medical AI systems. We sincerely hope this work encourages further exploration of calibration-aware training objectives in safety-critical domains beyond medical VQA. Acknowledgments. This work was supported by the National Natural Science Foundation of China under Grant 42394060 and 42394064, Ant Group Research Fund, and the Zhejiang University - Jolly Pharmaceutical Joint R&D Center for Intelligent Empowerment in Food and Medicine. Disclosure of Interests. The authors have no competing interests to declare.

References 1. Bai, S., et al.: Qwen2.5-vl technical report (2025), https://arxiv.org/abs/2502.13923 2. Chu, T., Zhai, Y., Yang, J., Tong, S., Xie, S., Schuurmans, D., Le, Q.V., Levine, S., Ma, Y.: Sft memorizes, rl generalizes: A comparative study of foundation model post-training (2025), https://arxiv.org/abs/2501.17161 3. DeepSeek-AI, et al.: Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning (2025), https://arxiv.org/abs/2501.12948 4. Deria, A., Kumar, K., Dukre, A.M., Segal, E., Khan, S., Razzak, I.: Medmo: Grounding and understanding multimodal large language model for medical images (2026), https://arxiv.org/abs/2602.06965 5. Du, Y., Wang, Y., Kong, M., Liang, T., Long, Q., Chen, B., Zhu, Q.: Confidence calibration for multimodal llms: An empirical study through medical vqa. In: proceedings of Medical Image Computing and Computer Assisted Intervention – MICCAI 2025. vol. LNCS 15965. Springer Nature Switzerland (September 2025) 6. Guo, C., Pleiss, G., Sun, Y., et al.: On calibration of modern neural networks. In: International Conference on Machine Learning. pp. 1321–1330. PMLR (2017) 7. He, X., Zhang, Y., Mou, L., Xing, E., Xie, P.: Pathvqa: 30000+ questions for medical visual question answering (2020), https://arxiv.org/abs/2003.10286 8. Huang, X., Wu, J., Liu, H., Tang, X., Zhou, Y.: Medvlthinker: Simple baselines for multimodal medical reasoning (2025), https://arxiv.org/abs/2508.02669 9. Lai, Y., Zhong, J., Li, M., Zhao, S., Yang, X.: Med-r1: Reinforcement learning for generalizable medical reasoning in vision-language models (2025), https://arxiv. org/abs/2503.13939 10. Lambert, N., et al.: Tulu 3: Pushing frontiers in open language model post-training (2025), https://arxiv.org/abs/2411.15124 11. Lau, J., Gayen, S., Ben Abacha, A., et al.: A dataset of clinically generated visual questions and answers about radiology images. Scientific Data 5(180251) (2018) 12. Li, C., Wong, C., Zhang, S., et al.: LLaVA-Med: Training a large language-andvision assistant for biomedicine in one day. In: Advances in Neural Information Processing Systems. vol. 36 (2024)

10

Y. Du et al.

13. Li, Z.Z., et al.: From system 1 to system 2: A survey of reasoning large language models (2025), https://arxiv.org/abs/2502.17419 14. Liu, B., Zhan, L.M., Xu, L., et al.: SLAKE: A semantically-labeled knowledgeenhanced dataset for medical visual question answering. In: IEEE 18th International Symposium on Biomedical Imaging. pp. 1650–1654 (2021) 15. Pan, J., Liu, C., Wu, J., Liu, F., Zhu, J., Li, H.B., Chen, C., Cheng, O., Rueckert, D.: MedVLM-R1: Incentivizing medical reasoning capability of vision-language models (VLMs) via reinforcement learning. arXiv preprint (2025), https://arxiv. org/abs/2502.19634, arXiv:2502.19634 16. Shao, Z., et al.: Deepseekmath: Pushing the limits of mathematical reasoning in open language models (2024), https://arxiv.org/abs/2402.03300 17. Shu, Y., Liu, C., Chen, R., Li, D., Dai, B.: Fleming-vl: Towards universal medical visual reasoning with multimodal llms (2025), https://arxiv.org/abs/2511.00916 18. Sun, H., Jiang, Y., Lou, W., Zhang, Y., Li, W., Wang, L., Liu, M., Liu, L., Wang, X.: Chiron-o1: Igniting multimodal large language models towards generalizable medical reasoning via mentor-intern collaborative search (2025), https://arxiv.org/ abs/2506.16962 19. Team, L., Xu, W., Chan, H.P., Li, L., Aljunied, M., Yuan, R., Wang, J., Xiao, C., Chen, G., Liu, C., Li, Z., Sun, Y., Shen, J., Wang, C., Tan, J., Zhao, D., Xu, T., Zhang, H., Rong, Y.: Lingshu: A generalist foundation model for unified multimodal medical understanding and reasoning (2025), https://arxiv.org/abs/2506.07044 20. Tian, K., Mitchell, E., Zhou, A., et al.: Just ask for calibration: Strategies for eliciting calibrated confidence scores from language models fine-tuned with human feedback. arXiv preprint (2023), https://arxiv.org/abs/2305.14975, arXiv:2305.14975 21. Xiong, M., Hu, Z., Lu, X., et al.: Can LLMs express their uncertainty? An empirical evaluation of confidence elicitation in LLMs. arXiv preprint (2023), https://arxiv. org/abs/2306.13063, arXiv:2306.13063 22. Xu, F., et al.: Towards large reasoning models: A survey of reinforced reasoning with large language models (2025), https://arxiv.org/abs/2501.09686

Related documents

• Clinical Validation of AI-Based System for Continuous Remote Monitoring of Patient Severity - Experts' Opinion#898802
• When Medicine Gets Irritating: Interactional and Structural Triggers of Uncertainty in Family Medicine Residency Training.#1039846
• In defence of post-hoc explanations in medical AI#73334
• Explainable Uncertainty Estimation for Reliable Medical AI#784819
• Explainable Uncertainty Estimation for Reliable Medical AI#783229