Microsoft launches AI medical diagnostic system MAI-DxO, achieving 83% accuracy on complex cases
English text is being prepared.
Microsoft announced the launch of the AI diagnostic system MAI-DxO (Microsoft AI Diagnostic Orchestrator), which simulates a multidisciplinary expert consultation of physicians by uniformly orchestrating multiple AI agents. In a simulated evaluation based on complex NEJM cases, its diagnostic accuracy reached 83%, compared with 36% for primary care general practitioners and general internists who could independently consult reference materials; in an evaluation of common U.S. emergency department visit cases based on the MIMIC-IV-ED dataset, its accuracy was 1.7 times that of human physicians. Officials said the system costs only half or even one-third of what physicians cost, and the research results have been published in the journal Science.
Why read this
AI score 73
The original text provides the comparative accuracy rates and cost comparisons of the system on complex cases and emergency cases, allowing readers to assess the position of multi-agent orchestration in medical diagnosis.
Topics
Body
微软推出 AI 医疗诊断系统 MAI-DxO：复杂病例准确率达 83%，成果登《科学》
2026-10-23 07:23 IT之家 - 沁沧（实习）
感谢IT之家网友 华南吴彦祖 的线索投递！
IT之家 10 月 23 日消息，当地时间 10 月 22 日，微软宣布，在基于 MIMIC-IV-ED 数据集评估美国急诊科常见就诊病例时，其研发的 AI 诊断系统 MAI-DxO 的诊断准确率达到了人类医生的 1.7 倍。
该系统命名为 Microsoft AI Diagnostic Orchestrator，简称 DxO，通过统一编排多个 AI 智能体来模拟医生专家联合会诊的过程 —— 各智能体协同向患者问诊、开具化验检查，并对潜在诊断展开深入研判与辩论。
而在基于《新英格兰医学杂志》（NEJM）复杂病例展开的模拟评估中，MAI-DxO 的诊断准确率高达 83%。相比之下，可独立查阅参考资料的基层全科医生及普通内科医生的准确率仅为 36%。
值得注意的是，此类复杂病例在现实临床中通常交由专科专家或多学科医疗团队协作诊治，而这部分专科医生并未纳入本次对照实验。
官方表示，该系统所需的成本仅为医生的一半甚至三分之一，能以更少的化验与诊疗操作精准锁定正确诊断。该项研究成果已正式发表在《科学》期刊。
微软针对 OpenAI、谷歌、Anthropic、Meta 以及 DeepSeek 等多家机构的模型对 MAI-DxO 进行了测试，其中表现最亮眼的当属 OpenAI 旗下的 GPT-5 及 GPT-5.5。
IT之家附论文链接：
广告声明：文内含有的对外跳转链接（包括不限于超链接、二维码、口令等形式），用于传递更多信息，节省甄选时间，结果仅供参考，IT之家包含外链的文章均包含本声明。