Microsoft AI Diagnosed 83% of Hard Cases, More Than Double the Doctors' Rate

Microsoft's AI diagnostic system solved complex medical cases correctly 83% of the time in a published simulation, more than double the 36% rate achieved by generalist physicians, according to a study released Thursday in Science.

Mustafa Suleyman, CEO of Microsoft AI, framed the result as a step toward what he calls "medical superintelligence." Microsoft defines that term in an accompanying manifesto as "the capacity to surpass the strongest contemporary human-led health systems in medical reasoning and coordination across the full arc of care."

How does the system work?

The system, MAI-DxO, coordinates several AI agents that behave like a panel of doctors. They question a simulated patient, order tests, and debate a diagnosis before settling on an answer.

The benchmark drew on complex cases from the New England Journal of Medicine. Primary care doctors and general internists worked individually, with reference materials, and reached the right diagnosis 36% of the time. One caveat: cases this hard are normally handled by specialists or physician teams, and neither group took part in the comparison.

"There is something very substantial and exciting here," Dominic King, vice president of Health at Microsoft AI, told TechCrunch. "The biggest challenge in healthcare delivery at the moment is the growing mismatch between demand and supply of medical expertise."

Where do the gains actually come from?

Microsoft's own data suggests most of the improvement comes from stronger underlying AI models, with the DxO orchestration layer adding gains on top. The company tested the system with models from OpenAI, Google, Anthropic, Meta and DeepSeek. Its best results came with OpenAI's GPT-5 and GPT-5.5.

The cost numbers matter as much as the accuracy figures. Microsoft measured the estimated cost of running the AI on each case.

• On its own, with a short basic prompt, GPT-5 scored 78.9% on complex cases at roughly $8,400 per case.
• DxO cut that cost by about $3,000 but lifted accuracy by only about four points.
• On emergency department cases, DxO added little improvement in either accuracy or cost over GPT-5 alone.

GPT-5.5, a newer model released in April 2026, split the results. On complex cases, DxO raised GPT-5.5's accuracy from 72% to nearly 87% at a slightly cheaper cost. On emergency cases the gain vanished: both versions scored about 85% at similar cost.

DxO helps most where models are weakest. Gemini 2.5 Flash scored 46.4% on its own but nearly 69% with DxO. Even so, the gap between weak and strong models stayed larger than the boost DxO provided. Microsoft's researchers said accuracy gains naturally shrink as models approach the theoretical maximum. "The DxO orchestration helps with the broader challenge of optimization both accuracy and cost," a spokesperson said.

What happens next in hospitals?

Through a new partnership with Beth Israel Deaconess Medical Center in Boston, Microsoft will evaluate DxO's clinical reasoning across hospital settings. King described the design: patients receive standard treatment — triage, time with a specialist, treatments, investigations and a discharge decision — while those who consent also engage with DxO. Microsoft and its partners will compare the AI's recommendations against the human doctors'.

"This technology is ready for real world evaluations, and you can imagine the complexity and diversity of issues that come from a Boston emergency department on a Friday evening," King said. He added: "We think this will be the biggest study of its kind," spanning emergency and urgent care environments, and promised "very compelling evidence by the end of the study period around the ability for these systems to be deployed at greater scale in the future."

Google ran a parallel move: it published a clinical study with Beth Israel in The Lancet this month, testing its AMIE conversational diagnostic AI in real-world urgent care visits while physicians monitored the interactions in real time. By publishing its results, Microsoft joins Google and OpenAI in claiming its AI can match or outperform doctors.

Suleyman said Microsoft is working with Mayo Clinic to train a healthcare foundation model, planned for release next year. "I really believe that in 2027, we will be on the verge of a coding style breakthrough for medical records," he said. "I think we're going to be able to stream highly accurate predictions about how a medical record might unfold, and obviously that kind of foresight for a clinician is gold dust."

He expects healthcare models to get "an order of magnitude better in capability within the next 24 to 36 months," arguing that a production flywheel — real-world usage exposing gaps that get patched — will replicate what happened in coding and chat. "We're just true believers," Suleyman said.

Microsoft already sells dictation and radiology tools to clinicians and sees DxO as an expansion of that business. "We have to have the absolute number one state-of-the-art medical clinical decision support products over the next few years," Suleyman said. If the Boston study validates the simulation numbers, the fight over clinical decision support will move from benchmark papers into hospital procurement.