跳到正文
arXiv cs.AI· Yicheng Gao, Xiaolin Zhou, Yahan Li, Yue Zhao, Ruishan Liu·· 10 小时前AI 评分38

MedExAgent:训练 LLM 智能体在嘈杂临床环境中问诊、检查与诊断

MedExAgent: Training LLM Agents to Ask, Examine, and Diagnose in Noisy Clinical Environments

AI 导读

MedExAgent 将临床诊断形式化为包含问诊、工具调用式医学检查和下诊断三类动作的 POMDP,并引入 7 种患者噪声与 3 种检查噪声的系统噪声模型。该智能体采用两阶段训练:先基于 Calgary-Cambridge 问诊模型构造的合成对话做监督微调,再用 DAPO 优化兼顾诊断准确率、工具调用质量和检查成本(含费用与患者不适)的复合奖励。

正文 · 原文

View PDF HTML (experimental)

Abstract:Real-world clinical diagnosis is a complex process in which the doctor is required to obtain information from both interaction with the patient and conducting medical exams. Additionally, the doctor needs to adapt to different patient personas, as well as noisy and incomplete information that can happen at any time during the process. However, existing benchmarks for medical LLMs and methods for automatic diagnosis largely simplify this process by reducing it to single-turn question answering, noise-free conversations, or sequential exam making, etc., ignoring the interactive and uncertain nature of clinical diagnosis. In this paper, we aim to address this gap by formalizing clinical diagnosis as a Partially Observable Markov Decision Process (POMDP) with three action types: questioning the patient, ordering medical exams as tool calls, and issuing a diagnosis. We also introduce a systematic noise model comprising seven patient noise types and three exam noise types. Using our proposed environment, we train an effective diagnosis agent, \textbf{MedExAgent}, through a two-stage pipeline that first performs supervised finetuning on synthetic conversations structured after the Calgary-Cambridge model for clinical interviews, and then applies DAPO to optimize a composite reward capturing diagnostic accuracy, tool call quality, and exam cost including financial cost and patient discomfort. Through extensive experiments and ablation studies, we demonstrate that MedExAgent achieves diagnostic performance comparable to larger models while maintaining cost-efficient examination strategies.
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as: arXiv:2605.07058 [cs.CL]
  (or arXiv:2605.07058v2 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2605.07058

arXiv-issued DOI via DataCite

Submission history

From: Yicheng Gao [view email]
[v1] Fri, 8 May 2026 00:12:23 UTC (384 KB)
[v2] Tue, 17 Nov 2026 21:18:35 UTC (411 KB)

来源:arXiv cs.AI · arxiv.org