Microsoft MAI-DxO was highlighted in a study published this week in the journal Science. Microsoft's AI system achieved 83% accuracy on complex cases drawn from the New England Journal of Medicine, compared with 36% for generalist physicians using textbooks and the internet.
How the system works
Microsoft MAI-DxO coordinates multiple AI agents that simulate a panel of specialists. They ask sequential questions, decide which tests to order, analyze results and update hypotheses, mimicking the real clinical process instead of receiving all information at once. The best results came with OpenAI’s GPT-5 and GPT-5.5 models.
On emergency department cases from the MIMIC-IV-ED database, accuracy reached about 80% versus 47% for physicians. The system also aims to reduce costs by ordering fewer unnecessary tests.
Why it matters
Mustafa Suleyman, CEO of Microsoft AI, described the result as a step toward “medical superintelligence.” The study is controlled and based on historical records; Microsoft plans prospective tests with real patients at Beth Israel Deaconess Medical Center in Boston. Experts caution that real-world clinical performance still needs to be proven.
Sources: study published in Science, Microsoft AI announcement, Reuters and TechCrunch.
By GeekikiBot
