MedFact-R1：通过伪标签增强实现事实性医学推理
MedFact-R1: Towards Factual Medical Reasoning via Pseudo-Label Augmentation
AI总结：
MedFact-R1提出两阶段框架，结合伪标签监督微调与分组相对策略优化及四个事实奖励信号，在三个医学问答基准上实现事实准确率最高22.5%的绝对提升。
摘要：
确保事实一致性和可靠推理仍然是医学视觉语言模型面临的关键挑战。我们提出了MEDFACT-R1，一个两阶段框架，将外部知识 grounding 与强化学习相结合，以改进事实性医学推理。第一阶段使用伪标签监督微调（SFT）引入外部事实专业知识；第二阶段应用分组相对策略优化（GRPO），并配备四个定制的事实奖励信号，以鼓励自洽推理。在三个公开医学问答基准上，MEDFACT-R1 的事实准确率相比先前最先进方法实现了高达22.5%的绝对提升。消融研究强调了伪标签 SFT 冷启动的必要性，并验证了每个 GRPO 奖励的贡献，凸显了知识 grounding 与强化学习驱动推理之间的协同作用对于可信医学人工智能的重要性。代码已发布在 https://github.com/Garfieldgengliang/MEDFACT-R1。
英文摘要：
Ensuring factual consistency and reliable reasoning remains a critical challenge for medical vision-language models. We introduce MEDFACT-R1, a two-stage framework that integrates external knowledge grounding with reinforcement learning to improve the factual medical reasoning. The first stage uses pseudo-label supervised fine-tuning (SFT) to incorporate external factual expertise; while the second stage applies Group Relative Policy Optimization (GRPO) with four tailored factual reward signals to encourage self-consistent reasoning. Across three public medical QA benchmarks, MEDFACT-R1 delivers up to 22.5% absolute improvement in factual accuracy over previous state-of-the-art methods. Ablation studies highlight the necessity of pseudo-label SFT cold start and validate the contribution of each GRPO reward, underscoring the synergy between knowledge grounding and RL-driven reasoning for trustworthy medical AI. Codes are released at https://github.com/Garfieldgengliang/MEDFACT-R1.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
