arXiv cs.AI· Xilin Dang, Weilin Ruan, Xue Yang, Jinghao Wang, Xiaowei Hu, Jinpeng Li, Pheng-Ann Heng·· 4 小时前AI 评分41
MedZERO：通过受控知识积累实现开放式医疗推理的自进化智能体
MedZERO: Self-Evolving Agents for Open-Ended Medical Reasoning Through Controlled Knowledge Accumulation
AI 导读
MedZERO 是一个面向开放式医疗推理的自进化框架，由生成医学题目的 Examiner 与借助外部知识工具进行多轮推理的 Reasoner 组成，并采用受控知识积累机制维护临时探索性知识与持久知识。在五个公开医疗推理基准上，基于 4B 和 8B 规模基座模型的 MedZERO 在开放式评测中取得所有受评方法中最高的平均准确率，代码已开源，论文被 NIPS 2026 接收。
正文 · 原文
Abstract:Large language models (LLMs) have shown promise in medical question answering and clinical reasoning, yet their improvement remains constrained by static parametric knowledge and costly expert supervision. Self-evolving agents offer a promising alternative by enabling models to improve through iterative task generation and problem-solving. However, most existing self-evolving methods are designed for easily verifiable domains such as mathematics and coding, where solutions can be checked by exact answers or executable programs. Medical reasoning is fundamentally different: it is open-ended, knowledge-intensive, and often only partially verifiable. We present MedZERO, a self-evolving framework for open-ended medical reasoning. MedZERO couples an Examiner that generates frontier medical question-option pairs with a Reasoner that solves them through evidence-grounded multi-turn reasoning with external knowledge tools. To support reliable, continual improvement, MedZERO adopts controlled knowledge accumulation, which maintains temporary exploratory knowledge and curated persistent knowledge in reasoning. We evaluate MedZERO on five public medical reasoning benchmarks using 4B- and 8B-scale base models under open-ended evaluation. At both model scales, MedZERO achieves the highest average accuracy among the evaluated methods. Code available at this https URL.
| Comments: | accepted by NIPS 2026 |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.08327 [cs.AI] |
| (or arXiv:2610.08327v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.08327 arXiv-issued DOI via DataCite |
Submission history
From: Xilin Dang [view email]
[v1]
Tue, 6 Oct 2026 13:25:55 UTC (617 KB)
[v2]
Sun, 18 Oct 2026 15:41:20 UTC (633 KB)
[v3]
Mon, 26 Oct 2026 07:07:35 UTC (638 KB)
来源：arXiv cs.AI · arxiv.org