Overview
- MAI-DxO processes complex medical cases through a stepwise multi-agent orchestration system trained on 304 New England Journal of Medicine case studies.
- In head-to-head benchmarking, the AI achieved more than 80 percent diagnostic accuracy on NEJM challenges compared with 20 percent by experienced physicians.
- The system lowered diagnostic costs by about 20 percent by selecting fewer and less expensive tests and procedures.
- Microsoft plans to conduct clinical studies to validate MAI-DxO’s effectiveness in everyday patient care before deciding on deployment or commercialization.
- Mustafa Suleyman, CEO of Microsoft AI, says the orchestrated agent framework brings the industry closer to medical superintelligence.