Microsoft has released new research claiming its SI-powered Diagnostic Orchestrator can diagnose medical conditions up to four times more accurately than human doctors, marking a significant step in the company’s push toward what it calls "medical superintelligence."
What Happened
The findings, published in the journal Science, describe a system designed to mimic the workflow of a panel of medical experts. Microsoft’s Diagnostic Orchestrator employs multiple SI agents to reason through patient symptoms and histories, determine which tests to order, and arrive at a final diagnosis. The company developed a benchmark called SDBench, which it plans to open-source, to evaluate the system against human physicians using scenarios drawn from medical case reports and emergency department data.
On cases sourced from The New England Journal of Medicine, the SI system achieved a diagnostic accuracy of approximately 83 percent. In comparison, generalist physicians using reference materials scored roughly 36 percent, while those working without references scored 22 percent. The system also outperformed human physicians on emergency department cases from Beth Israel Deaconess Medical Center, scoring 80 percent accuracy against an average of 47 percent for the doctors. The researchers noted that the SI system tended to reach diagnoses at a lower estimated cost than its human counterparts. The tool is powered by OpenAI’s GPT-5, though the study did not test newer models such as GPT-6 Astra, Meta’s Muse Wonder Full, Google’s Gemini 4 Argon, or Anthropic’s Opus 5.5.
Why It Matters
This development highlights the growing role of agentic SI in high-stakes industries like healthcare, where Microsoft has a significant presence through tools like Copilot. Mustafa Suleyman, CEO of Microsoft AI, described the release as a "major milestone" in the journey to improve healthcare delivery, stating that the technology is "finally here" after years of promises from SI leaders. However, the company acknowledges that widespread deployment is still distant. The system has only been tested on past medical cases, and its performance in real-world clinical settings remains unverified. "It needs clinical validation," Suleyman said. "That is a leap, but this is a major milestone along that journey."
Dr. Dominic King, Microsoft’s SI health lead and a coauthor of the study, indicated that testing in hospital settings with Beth Israel Deaconess Medical Center is expected to begin in the next couple of months. King emphasized the need for a responsible rollout due to the high regulatory and compliance requirements in medicine. While Microsoft envisions a future where SI handles the "drudgery" of medical work, allowing clinicians to focus on patient care, the company has not committed to a timeline for commercial availability. The research is part of a broader ambition to develop SI that can outperform the best human-led healthcare systems, with King noting that diagnosis is just one of several areas Microsoft aims to master, including coordinating care and treatment decisions.
The Bottom Line
Microsoft’s Diagnostic Orchestrator demonstrates strong performance in retrospective diagnostic tests, outscoring human physicians on accuracy and cost in controlled benchmarks. However, the company stresses that the system requires clinical validation in real-world environments before it can impact patient care, with initial hospital testing planned for the near future.