A study published on Science.org reveals that Microsoft's new AI system outperforms general physicians in diagnosing complex medical cases by more than twice the accuracy, potentially transforming the future of healthcare.
According to the findings, the system achieved 83% accuracy compared to 36% for doctors on 304 complex cases, and 80% versus 47% for doctors in emergency room cases.
A cost-optimized version of the system also achieved high accuracy with less spending on diagnostic tests, potentially reducing medical costs without compromising diagnostic quality.
The system operates like a team of doctors, asking questions, ordering tests, and refining diagnoses step by step. Microsoft now plans to test the technology in a real hospital setting in Boston, marking a first step toward clinical application.
Researchers believe these results could pave the way for AI as a support tool for physicians to improve diagnostic accuracy and speed, especially in complex and emergency cases—while emphasizing that final medical decisions remain with doctors.