MedCEG: Reinforcing Verifiable Medical Reasoning with Critical Evidence Graph

2025-12-15

AI Summary

Clinical large language models lack rigorous clinical validation of their reasoning processes, hindering their trustworthy deployment in clinical decision support. To address this, we propose the Clinical Evidence Graph (CEG)—a verifiable and traceable structured reasoning-path modeling framework that explicitly constrains medical reasoning to evidence-based nodes and causal chains for the first time. Methodologically, we design a three-dimensional reward function measuring node coverage, structural correctness, and chain completeness, integrated with clinical-knowledge-guided Proximal Policy Optimization (PPO) reinforcement learning and an algorithmic CEG generation pipeline. Our approach achieves significant improvements over state-of-the-art methods across multiple medical reasoning benchmarks. Generated reasoning chains receive high clinical credibility scores from domain experts (mean 4.82/5.0). We publicly release our code, models, and a challenging case dataset to foster reproducible research and clinical evaluation.

Technology Category

• Reasoning under Uncertainty: Graphical Models
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language models with reasoning capabilities have demonstrated impressive performance across a wide range of domains. In clinical applications, a transparent, step-by-step reasoning process provides physicians with strong evidence to support decision-making. While reinforcement learning has effectively enhanced reasoning performance in medical contexts, the clinical reliability of these reasoning processes remains limited because their accuracy and validity are often overlooked during training. To address this gap, we propose MedCEG, a framework that augments medical language models with clinically valid reasoning pathways by explicitly supervising the reasoning process through a Critical Evidence Graph (CEG). We curate a dataset of challenging clinical cases and algorithmically construct a CEG for each sample to represent a high-quality verifiable reasoning pathway. To guide the reasoning process, we introduce a Clinical Reasoning Procedure Reward, which evaluates Node Coverage, Structural Correctness, and Chain Completeness, thereby providing a holistic assessment of reasoning quality. Experimental results show that MedCEG surpasses existing methods in performance while producing clinically valid reasoning chains, representing a solid advancement in reliable medical AI reasoning. The code and models are available at https://github.com/LinjieMu/MedCEG.

Problem

• Enhances medical AI reasoning with verifiable evidence graphs
• Addresses limited clinical reliability in reinforcement learning models
• Improves reasoning quality through structured clinical validation metrics

Innovation

• Uses Critical Evidence Graph for verifiable reasoning pathways
• Introduces Clinical Reasoning Procedure Reward for quality assessment
• Enhances medical language models with clinically valid reasoning