跳转至

文章背景与核心概要

大语言模型（LLM）在医学诊断领域展现出了巨大的应用前景，但现有的提示词工程和多智能体系统通常局限于孤立的单次推理，缺乏对可复用临床经验的积累。为了填补这一空白，本文提出了 MACD（Multi-Agent Clinical Diagnosis） 框架，通过结构化的多智能体流水线让大模型实现临床知识的自学习——像人类医师一样总结、提炼并应用诊疗经验。

此外，作者还提出了一套 MACD-human 协同工作流，将大模型迭代会诊、裁判智能体以及人类监督有机结合。在全新构建的 MIMIC-MACD 队列（包含涵盖七种疾病的 4,390 个真实世界病例）上的评估表明，MACD 显著提升了开源大模型的初诊准确率，表现超越了权威知识库，并展现出强大的“人机协同”潜力。

MACD: Multi-Agent Clinical Diagnosis with Self-Learned Knowledge for LLM

arXiv ID: 2509.20067
Primary Subject: Artificial Intelligence (cs.AI)
Authors: Wenliang Li, Rui Yan, Xu Zhang, Li Chen, Hongji Zhu, Jing Zhao, Junjun Li, Mengru Li, Wei Cao, Zihang Jiang, Wei Wei, Kun Zhang, Shaohua Kevin Zhou
Timeline: Submitted on September 24, 2025; Last revised on August 24, 2026 (v5)

📋 总结 (Summary)

Large Language Models (LLMs) show immense promise in medical diagnosis, yet current prompting and multi-agent systems typically focus on isolated inferences without accumulating reusable clinical experience. To bridge this gap, this paper introduces MACD (Multi-Agent Clinical Diagnosis), a framework enabling LLMs to self-learn clinical knowledge through a structured multi-agent pipeline—summarizing, refining, and applying insights much like human physicians.

Additionally, the authors propose a MACD-human collaborative workflow combining iterative LLM consultations, a judge agent, and human oversight. Evaluated on the newly constructed MIMIC-MACD cohort (4,390 real-world cases across seven diseases), MACD significantly enhances open-weight LLMs' primary diagnostic accuracy, outperforming authoritative knowledge bases and demonstrating powerful human-AI synergy.

大语言模型（LLM）在辅助医学诊断方面显示出了广阔的前景，基于提示词的方法为其能力提升提供了一种灵活且可部署的途径。然而，现有的提示工程和多智能体方法往往侧重于优化单一推理，较少关注从临床实践中积累可复用的经验，从而限制了它们的现实适用性。

为了解决这一问题，本研究提出了一种新颖的多智能体临床诊断（MACD）框架，该框架允许大模型通过多智能体流水线自主学习临床知识，实现诊断见解的总结、提炼和应用，从而模仿人类医师的专业成长过程。我们进一步将其扩展到MACD-human协同工作流中，其中多个基于LLM的诊断智能体参与迭代会诊，并在未能达成一致的病例中由裁判智能体和人类监督提供支持。

研究所构建的MIMIC-MACD队列包含来自7种疾病的4,390个真实世界患者病例，其中包括1,314个用于知识学习的病例和3,076个用于评估的保留病例。在各种开源大模型中，MACD显著提升了初诊准确率，相比已建立的权威知识平均提升了11.6个百分点，同时缩小了开源模型与顶尖专有大模型之间的性能差距。此外，MACD-human工作流在纯文本病例 vignettes（病例小品）上的表现比纯医师诊断提升了18.3个百分点，证明了人机协同的协同潜力。这项工作因此提出了一个可扩展的自学习范式，弥合了LLM内在知识与现实临床实践需求之间的鸿沟，向着可靠、可解释且可部署的AI辅助诊断迈出了重要一步。

🧠 摘要 (Abstract)

Large language models (LLMs) have shown promise in supporting medical diagnosis, with prompting-based methods offering a flexible and deployable means of capability enhancement. However, existing prompt engineering and multi-agent approaches often focus on optimizing single inferences, paying less attention to the accumulation of reusable experience from clinical practice, constraining their real-world applicability.

To address this, this study proposes a novel Multi-Agent Clinical Diagnosis (MACD) framework, which allows LLMs to self-learn clinical knowledge via a multi-agent pipeline that summarizes, refines, and applies diagnostic insights, mirroring the professional development of human physicians. We further extend it to a MACD-human collaborative workflow, where multiple LLM-based diagnostician agents engage in iterative consultations, supported by a judge agent and human oversight for cases where agreement is not reached.

The MIMIC-MACD cohort comprising 4,390 real-world patient cases across seven diseases is constructed, including 1,314 cases for knowledge learning and 3,076 held-out cases for evaluation. Across diverse open-weight LLMs, MACD significantly improves primary diagnostic accuracy, achieving an average improvement of 11.6 percentage points over established authoritative knowledge, while narrowing the performance gap between open-weight models and state-of-the-art LLMs. Furthermore, the MACD-human workflow yields an 18.3-percentage-point improvement over physician-only diagnosis on text-only vignettes, demonstrating the synergistic potential of human-AI collaboration. This work thus presents a scalable self-learning paradigm that bridges the gap between the intrinsic knowledge of LLMs and the demands of real-world clinical practice, advancing towards a reliable, interpretable, and deployable AI-assisted diagnosis.

大语言模型（LLM）在辅助医学诊断方面显示出了广阔的前景，基于提示词的方法为其能力提升提供了一种灵活且可部署的途径。然而，现有的提示工程和多智能体方法往往侧重于优化单一推理，较少关注从临床实践中积累可复用的经验，从而限制了它们的现实适用性。

为了解决这一问题，本研究提出了一种新颖的多智能体临床诊断（MACD）框架，该框架允许大模型通过多智能体流水线自主学习临床知识，实现诊断见解的总结、提炼和应用，从而模仿人类医师的专业成长过程。我们进一步将其扩展到MACD-human协同工作流中，其中多个基于LLM的诊断智能体参与迭代会诊，并在未能达成一致的病例中由裁判智能体和人类监督提供支持。

研究所构建的MIMIC-MACD队列包含来自7种疾病的4,390个真实世界患者病例，其中包括1,314个用于知识学习的病例和3,076个用于评估的保留病例。在各种开源大模型中，MACD显著提升了初诊准确率，相比已建立的权威知识平均提升了11.6个百分点，同时缩小了开源模型与顶尖专有大模型之间的性能差距。此外，MACD-human工作流在纯文本病例 vignettes（病例小品）上的表现比纯医师诊断提升了18.3个百分点，证明了人机协同的协同潜力。这项工作因此提出了一个可扩展的自学习范式，弥合了LLM内在知识与现实临床实践需求之间的鸿沟，向着可靠、可解释且可部署的AI辅助诊断迈出了重要一步。

(License Reference: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International)
license icon