Microsoft Says Its AI Beats Physicians At Diagnosis
Microsoft AI published a post on October 22, 2026 titled "Building Medical Superintelligence," alongside a peer-reviewed paper in Science. Everything below comes from that first-party post.
Transcript
Microsoft says its AI beats physicians at diagnosis. The research is now peer reviewed.
The journal Science published the study this week. Microsoft's AI system runs a panel of virtual physicians who debate before agreeing on a diagnosis.
On complex New England Journal of Medicine cases, the system reached 83 percent accuracy. Physicians using reference materials scored 36 percent.
On ordinary emergency department cases it was 1.7 times more accurate than physicians. Microsoft reports lower costs from ordering fewer tests.
Microsoft graded its own system on recorded cases, not live patients. The hospital trial has not started, and part of the benchmark is public.
Inboxsmith builds AI receptionists for small business. Please like and subscribe for more news.
Sources
Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:
- Our work is published today in Science.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- Today's paper, building on earlier work, introduces two foundations for that shift: the Sequential Diagnosis Benchmark (SDBench), which tests step-by-step diagnostic reasoning, and the Microsoft AI Diagnostic Orchestrator (MAI-DxO), our leading AI system on the benchmark.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- Rather than relying on one model for an immediate answer, it operates like a panel of virtual physicians: each proposing diagnoses, selecting useful questions or tests, and challenging the emerging view.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- MAI-DxO coordinates this exchange over several rounds of debate, updating its assessment as evidence arrives and weighing each test's likely value against its cost.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- On complex NEJM cases, MAI-DxO achieved 83% diagnostic accuracy. Physicians achieved 22% without external reference resources and 36% with them, making MAI-DxO about 3.7 and 2.3 times more accurate, respectively.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- On more typical emergency-department cases from MIMIC-IV-ED, MAI-DxO achieved 80% accuracy versus 47% for physicians, more than 1.7 times higher.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- It also manages to do this at half or even one third of the cost, getting to the right diagnosis with fewer tests and procedures.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- SDBench evaluates performance across two complementary sets of cases: 304 complex cases taken from The New England Journal of Medicine (NEJM) case proceedings and a set of more typical emergency-department presentations from MIMIC-IV-ED.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- We are open-sourcing SDBench for select NEJM cases, including the benchmark, Gatekeeper, Judge and cost-estimation code, so researchers can reproduce the evaluation, test other models, and challenge or extend the findings.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)
- Through a new partnership with Beth Israel Deaconess Medical Center in Boston, we will evaluate clinical reasoning capabilities across different hospital settings. In a prospective study, patients will interact with the system directly, with performance measured against physicians' independent assessments.(Microsoft AI (microsoft.ai), first party announcement titled "Building Medical Superintelligence", MAI Health Team, October 22, 2026)