微软于2026年10月22日星期四公布了发表在《科学》期刊上的研究结果，表明其诊断编排系统在医学病例报告上的诊断准确率高达百分之83，相比之下，使用参考资料的全科医生准确率为百分之36。
Microsoft Unveils GPT-5-Powered Diagnostic Orchestrator Outperforming Doctors in Study Published in Science On Thursday, October 22, 2026, Microsoft revealed findings published in the journal Science demonstrating that its Diagnostic Orchestrator system achieved diagnostic accuracy rates of up to 83 percent on medical case reports, compared with 36 percent for generalist physicians using references.
该系统由OpenAI的GPT-5驱动，部署了多个人工智能代理，模拟由医学专家组成的小组，以推理症状、选择检查并以较低的估计成本制定诊断。
Powered by OpenAI's GPT-5, the system deploys multiple AI agents that simulate a panel of medical experts to reason through symptoms, select tests, and formulate diagnoses at a lower estimated cost.
微软人工智能首席执行官穆斯塔法·苏莱曼和医疗负责人多米尼克·金博士表示，该工具将在未来几个月内与波士顿的贝丝以色列女执事医疗中心合作进行现实世界的临床测试，不过广泛的商业推广仍有待临床验证和监管合规。
Microsoft AI CEO Mustafa Suleyman and health lead Dr. Dominic King stated that the tool will undergo real-world clinical testing with Boston's Beth Israel Deaconess Medical Center in coming months, though widespread commercial rollout remains pending clinical validation and regulatory compliance.