Microsoft's MAI-DxO AI diagnostic system outperforms physicians on complex cases in Science study; Microsoft eyes 'medical superintelligence'
Microsoft AI published a study in Science showing its MAI-DxO diagnostic system, which coordinates several AI agents acting as a panel of doctors, correctly diagnosed 83% of complex New England Journal of Medicine cases in simulation. Primary care doctors and general internists working alone scored 36%, though specialists and physician teams, who normally handle such cases, were not part of the comparison. Mustafa Suleyman, CEO of Microsoft AI, framed the result as a step toward "medical superintelligence". This is product and research news with no direct threat or security impact.
Detalhes técnicos
- The study's own data suggests most of the gain comes from stronger underlying models, with the orchestration layer adding smaller improvements on top.
- GPT-5 alone scored 78.9% on complex cases at about $8,400 per case. DxO cut the cost by about $3,000 but raised accuracy by only about four points.
- GPT-5.5 rose from 72% to nearly 87% with DxO on complex cases, at a slightly lower cost. Emergency department cases showed little or no gain, with both setups scoring about 85%.
- Gemini 2.5 Flash rose from 46.4% to nearly 69% with DxO.
- Microsoft is working with the Mayo Clinic on a healthcare foundation model planned for release next year. It is also starting a real-world evaluation with Beth Israel Deaconess Medical Center in Boston, in which consenting patients will use DxO in parallel with standard care.
Linha do tempo
-
Microsoft's MAI-DxO AI diagnostic system outperforms physicians on complex cases in Science study; Microsoft eyes 'medical superintelligence'
ler a análise desta fonte ocultar
Microsoft AI published a study in Science showing its MAI-DxO diagnostic system, which coordinates several AI agents acting as a panel of doctors, correctly diagnosed 83% of complex New England Journal of Medicine cases in simulation. Primary care doctors and general internists working alone scored 36%, though specialists and physician teams, who normally handle such cases, were not part of the comparison. Mustafa Suleyman, CEO of Microsoft AI, framed the result as a step toward "medical superintelligence". This is product and research news with no direct threat or security impact.
Technical Details
- The study's own data suggests most of the gain comes from stronger underlying models, with the orchestration layer adding smaller improvements on top.
- GPT-5 alone scored 78.9% on complex cases at about $8,400 per case. DxO cut the cost by about $3,000 but raised accuracy by only about four points.
- GPT-5.5 rose from 72% to nearly 87% with DxO on complex cases, at a slightly lower cost. Emergency department cases showed little or no gain, with both setups scoring about 85%.
- Gemini 2.5 Flash rose from 46.4% to nearly 69% with DxO.
- Microsoft is working with the Mayo Clinic on a healthcare foundation model planned for release next year. It is also starting a real-world evaluation with Beth Israel Deaconess Medical Center in Boston, in which consenting patients will use DxO in parallel with standard care.