A new Science study says Microsoft's AI diagnostic tool more than doubled general physicians' accuracy on complex cases.
A Microsoft AI diagnostic tool that works through patients step by step, as doctors do, was more than twice as accurate as general physicians at solving complex medical cases, according to a study published Thursday, in Science.
At a glance: Microsoft AI researchers published the findings Thursday, October 22, 2026. On more than 300 unusual cases, the tool outperformed general physicians who had textbooks and internet access. Microsoft's next step is a triage test in emergency department waiting rooms.
How does the Microsoft AI diagnostic tool work?
Most existing medical AI takes an all-at-once approach, examining symptoms, lab results and imaging together to produce a diagnosis. One pass, one answer.
Microsoft's tool works differently. An orchestrator decides what questions to ask, which examinations and tests to order and whether further testing is needed, updating its diagnostic hypotheses as it goes.
How accurate was it?
On more than 300 highly unusual and confusing cases, the model using the orchestrator was more than twice as accurate as general physicians using textbooks and the internet. It also outperformed physicians on scenarios drawn from a major database of emergency room admissions and the work-ups it designed cost less than those ordered by human doctors.
The researchers say older systems are usually graded on multiple-choice questions. "AI has become very good at getting a 100% pass rate on those," said Dr. Dominic King, the corporate vice president of the health unit at Microsoft AI.
Is the tool ready for hospitals?
Not yet. Microsoft AI is working with Beth Israel Deaconess Medical Center in Boston to test whether the tool can triage patients in emergency department waiting rooms.
Caveats from the 2025 version of this research are worth keeping in view. That work used simulated, text-only interactions rather than real patients and was not measured against patient outcomes. Its benchmark cases came from complex, teaching-focused journal material and the comparison group was generalist physicians, not specialists. The Science paper itself wasn't available for this report, so whether those limits still apply is unconfirmed.
Microsoft has said it sees the technology as a "complement to doctors and other health professionals."
Microsoft's results suggest that having AI reason through a case in stages, the way clinicians do, may beat the single-pass approach on hard diagnostic puzzles. The emergency room test with Beth Israel Deaconess will show whether that holds up outside a controlled benchmark.
Until then, Business Fortune believes this is research, not a clinical tool.
FAQs
What is the Microsoft AI diagnostic tool?
It's a system that diagnoses patients step by step, using an orchestrator to choose questions and tests as it refines its hypotheses.
How much more accurate was Microsoft's AI than doctors?
Reuters reports it was more than twice as accurate as general physicians using textbooks and the internet on 300-plus complex cases.
Where was the Microsoft AI diagnostic study published?
In Science, on Thursday, October 22, 2026.
Can I use Microsoft's AI diagnostic tool now?
No release has been announced. Microsoft is first testing it for emergency room triage with Beth Israel Deaconess Medical Center.
Will AI replace doctors?
Microsoft says it views the technology as a complement to doctors and other health professionals.
Sources
Reuters via Detroit News, Reuters via Yahoo Finance Canada, Microsoft AI, Fortune, MobiHealthNews, ScienceBlog.















Comments