MedS$^3$: Towards Medical Slow Thinking with Self-Evolved Soft Dual-sided Process Supervision

• URL: http://arxiv.org/abs/2501.12051v3
• Date: Wed, 10 Sep 2025 02:53:11 GMT
• Title: MedS$^3$: Towards Medical Slow Thinking with Self-Evolved Soft Dual-sided Process Supervision
• Authors: Shuyang Jiang, Yusheng Liao, Zhe Chen, Ya Zhang, Yanfeng Wang, Yu Wang,
• Abstract summary: mone is a self-evolving framework that imparts robust reasoning capabilities to small, deployable models.<n>mone outperforms the previous state-of-the-art medical model by +6.45 accuracy points and surpasses 32B-scale general-purpose reasoning models by +8.57 points.
• Score: 42.03114317779815
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Medical language models face critical barriers to real-world clinical reasoning applications. However, mainstream efforts, which fall short in task coverage, lack fine-grained supervision for intermediate reasoning steps, and rely on proprietary systems, are still far from a versatile, credible and efficient language model for clinical reasoning usage. To this end, we propose \mone, a self-evolving framework that imparts robust reasoning capabilities to small, deployable models. Starting with 8,000 curated instances sampled via a curriculum strategy across five medical domains and 16 datasets, we use a small base policy model to conduct Monte Carlo Tree Search (MCTS) for constructing rule-verifiable reasoning trajectories. Self-explored reasoning trajectories ranked by node values are used to bootstrap the policy model via reinforcement fine-tuning and preference learning. Moreover, we introduce a soft dual process reward model that incorporates value dynamics: steps that degrade node value are penalized, enabling fine-grained identification of reasoning errors even when the final answer is correct. Experiments on eleven benchmarks show that \mone outperforms the previous state-of-the-art medical model by +6.45 accuracy points and surpasses 32B-scale general-purpose reasoning models by +8.57 points. Additional empirical analysis further demonstrates that \mone achieves robust and faithful reasoning behavior.

Related papers

• Evaluating and Enhancing the Vulnerability Reasoning Capabilities of Large Language Models [15.849480549367684]
  We propose DAGVul, a novel framework that models vulnerability reasoning as a Directed Acyclic Graph (DAG) generation task.<n>By further introducing Reinforcement Learning with Verifiable Rewards (RLVR), we align model reasoning trace with program-intrinsic logic.<n>Our framework improves the reasoning F1-score by an average of 18.9% over all the baselines.
  arXiv  Detail & Related papers  (2026-02-06T13:19:45Z)
• Pushing the Boundaries of Natural Reasoning: Interleaved Bonus from Formal-Logic Verification [49.506412445511934]
  Large Language Models (LLMs) show remarkable capabilities, yet their next-token prediction creates logical inconsistencies and reward hacking.<n>We introduce a formal logic verification-guided framework that dynamically interleaves formal symbolic verification with the natural language generation process.<n>We operationalize this framework via a novel two-stage training pipeline that synergizes formal logic verification-guided supervised fine-tuning and policy optimization.
  arXiv  Detail & Related papers  (2026-01-30T07:01:25Z)
• Catch Me If You Can: How Smaller Reasoning Models Pretend to Reason with Mathematical Fidelity [15.774418410083515]
  We introduce a diagnostic framework that distinguishes genuine mathematical reasoning from superficial pattern matching.<n>We reveal a striking disconnect between surface performance and reasoning fidelity.<n>Our diagnostics expose reasoning failures invisible to traditional accuracy metrics.
  arXiv  Detail & Related papers  (2025-11-29T16:47:01Z)
• Med-PRM: Medical Reasoning Models with Stepwise, Guideline-verified Process Rewards [21.831262938278915]
  We introduce Med-PRM, a process reward modeling framework to verify each reasoning step against established medical knowledge bases.<n>Med-PRM achieves state-of-the-art performance, with improving the performance of base models by up to 13.50%.<n>We demonstrate the generality of Med-PRM by integrating it in a plug-and-play fashion with strong policy models such as Meerkat.
  arXiv  Detail & Related papers  (2025-06-13T05:36:30Z)
• ReasonMed: A 370K Multi-Agent Generated Dataset for Advancing Medical Reasoning [54.30630356786752]
  ReasonMed is the largest medical reasoning dataset to date, with 370k high-quality examples.<n>It is built through a multi-agent generation, verification, and refinement process.<n>Using ReasonMed, we find that integrating detailed CoT reasoning with concise answer summaries yields the most robust fine-tuning results.
  arXiv  Detail & Related papers  (2025-06-11T08:36:55Z)
• Disentangling Reasoning and Knowledge in Medical Large Language Models [23.401484250342158]
  Medical reasoning in large language models aims to emulate clinicians' diagnostic thinking.<n>Current benchmarks such as MedQA-USMLE, MedMCQA, and PubMedQA often mix reasoning with factual recall.<n>We evaluate biomedical models (HuatuoGPT-o1, MedReason, m1) and general-domain models (DeepSeek-R1, o4-mini, Qwen3)<n>We train BioMed-R1 using fine-tuning and reinforcement learning on reasoning-heavy examples.
  arXiv  Detail & Related papers  (2025-05-16T17:16:27Z)
• Benchmarking Chinese Medical LLMs: A Medbench-based Analysis of Performance Gaps and Hierarchical Optimization Strategies [11.0505830548286]
  This study introduces a granular error taxonomy through systematic analysis of top 10 models on MedBench.<n> Evaluation of 10 leading models reveals vulnerabilities, despite achieving 0.86 accuracy in medical knowledge recall.<n>Our analysis uncovers systemic weaknesses in knowledge boundary enforcement and multi-step reasoning.
  arXiv  Detail & Related papers  (2025-03-10T13:28:25Z)
• Benchmarking Reasoning Robustness in Large Language Models [76.79744000300363]
  We find significant performance degradation on novel or incomplete data.<n>These findings highlight the reliance on recall over rigorous logical inference.<n>This paper introduces a novel benchmark, termed as Math-RoB, that exploits hallucinations triggered by missing information to expose reasoning gaps.
  arXiv  Detail & Related papers  (2025-03-06T15:36:06Z)
• Causality can systematically address the monsters under the bench(marks) [64.36592889550431]
  Benchmarks are plagued by various biases, artifacts, or leakage.<n>Models may behave unreliably due to poorly explored failure modes.<n> causality offers an ideal framework to systematically address these challenges.
  arXiv  Detail & Related papers  (2025-02-07T17:01:37Z)
• Towards Evaluating and Building Versatile Large Language Models for Medicine [57.49547766838095]
  We present MedS-Bench, a benchmark designed to evaluate the performance of large language models (LLMs) in clinical contexts. MedS-Bench spans 11 high-level clinical tasks, including clinical report summarization, treatment recommendations, diagnosis, named entity recognition, and medical concept explanation. MedS-Ins comprises 58 medically oriented language corpora, totaling 13.5 million samples across 122 tasks.
  arXiv  Detail & Related papers  (2024-08-22T17:01:34Z)
• Is larger always better? Evaluating and prompting large language models for non-generative medical tasks [11.799956298563844]
  This study benchmarks various models, including GPT-based LLMs, BERT-based models, and traditional clinical predictive models. We focused on tasks such as readmission and prediction, disease hierarchy reconstruction, and biomedical sentence matching. Results indicated that LLMs exhibited robust zero-shot predictive capabilities on structured EHR data when using well-designed prompting strategies. For unstructured medical texts, LLMs did not outperform finetuned BERT models, which excelled in both supervised and unsupervised tasks.
  arXiv  Detail & Related papers  (2024-07-26T06:09:10Z)
• Medical Vision-Language Pre-Training for Brain Abnormalities [96.1408455065347]
  We show how to automatically collect medical image-text aligned data for pretraining from public resources such as PubMed. In particular, we present a pipeline that streamlines the pre-training process by initially collecting a large brain image-text dataset. We also investigate the unique challenge of mapping subfigures to subcaptions in the medical domain.
  arXiv  Detail & Related papers  (2024-04-27T05:03:42Z)
• Towards a clinically accessible radiology foundation model: open-access and lightweight, with automated evaluation [113.5002649181103]
  Training open-source small multimodal models (SMMs) to bridge competency gaps for unmet clinical needs in radiology. For training, we assemble a large dataset of over 697 thousand radiology image-text pairs. For evaluation, we propose CheXprompt, a GPT-4-based metric for factuality evaluation, and demonstrate its parity with expert evaluation. The inference of LlaVA-Rad is fast and can be performed on a single V100 GPU in private settings, offering a promising state-of-the-art tool for real-world clinical applications.
  arXiv  Detail & Related papers  (2024-03-12T18:12:02Z)
• Learnable Weight Initialization for Volumetric Medical Image Segmentation [66.3030435676252]
  We propose a learnable weight-based hybrid medical image segmentation approach. Our approach is easy to integrate into any hybrid model and requires no external training data. Experiments on multi-organ and lung cancer segmentation tasks demonstrate the effectiveness of our approach.
  arXiv  Detail & Related papers  (2023-06-15T17:55:05Z)
• Federated Learning of Medical Concepts Embedding using BEHRT [0.0]
  We propose a federated learning approach for learning medical concepts embedding. Our approach is based on embedding model like BEHRT, a deep neural sequence model for EHR. We compare the performance of a model trained with FL against a model trained on centralized data.
  arXiv  Detail & Related papers  (2023-05-22T14:05:39Z)
• Do We Still Need Clinical Language Models? [15.023633270864675]
  We show that relatively small specialized clinical models substantially outperform all in-context learning approaches. We release the code and the models used under the PhysioNet Credentialed Health Data license and data use agreement.
  arXiv  Detail & Related papers  (2023-02-16T05:08:34Z)
• Large Language Models Encode Clinical Knowledge [21.630872464930587]
  Large language models (LLMs) have demonstrated impressive capabilities in natural language understanding and generation. We propose a framework for human evaluation of model answers along multiple axes including factuality, precision, possible harm, and bias. We show that comprehension, recall of knowledge, and medical reasoning improve with model scale and instruction prompt tuning.
  arXiv  Detail & Related papers  (2022-12-26T14:28:24Z)
• Competence-based Multimodal Curriculum Learning for Medical Report Generation [98.10763792453925]
  We propose a Competence-based Multimodal Curriculum Learning framework ( CMCL) to alleviate the data bias and make best use of available data. Specifically, CMCL simulates the learning process of radiologists and optimize the model in a step by step manner. Experiments on the public IU-Xray and MIMIC-CXR datasets show that CMCL can be incorporated into existing models to improve their performance.
  arXiv  Detail & Related papers  (2022-06-24T08:16:01Z)
• A multi-stage machine learning model on diagnosis of esophageal manometry [50.591267188664666]
  The framework includes deep-learning models at the swallow-level stage and feature-based machine learning models at the study-level stage. This is the first artificial-intelligence-style model to automatically predict CC diagnosis of HRM study from raw multi-swallow data.
  arXiv  Detail & Related papers  (2021-06-25T20:09:23Z)
• On the model-based stochastic value gradient for continuous reinforcement learning [50.085645237597056]
  We show that simple model-based agents can outperform state-of-the-art model-free agents in terms of both sample-efficiency and final reward. Our findings suggest that model-based policy evaluation deserves closer attention.
  arXiv  Detail & Related papers  (2020-08-28T17:58:29Z)
• Med7: a transferable clinical natural language processing model for electronic health records [6.935142529928062]
  We introduce a named-entity recognition model for clinical natural language processing. The model is trained to recognise seven categories: drug names, route, frequency, dosage, strength, form, duration. We evaluate the transferability of the developed model using the data from the Intensive Care Unit in the US to secondary care mental health records (CRIS) in the UK.
  arXiv  Detail & Related papers  (2020-03-03T00:55:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.