Microsoft's artificial intelligence-powered medical diagnosis system achieved 2.3 times higher diagnostic accuracy than physicians' benchmark results in tests involving historical medical cases, according to a new study.
The US-based technology company evaluated its AI diagnostic system, known as MAI-DxO, using 304 complex medical cases previously published in The New England Journal of Medicine.
According to Microsoft, the system achieved approximately 2.3 times higher diagnostic accuracy than doctors who had access to external resources, including medical books and online information.
When compared with doctors who attempted to diagnose cases without consulting external sources, the AI system performed approximately 3.7 times better.
The findings, published in the journal Science, highlight the potential of artificial intelligence to support medical diagnosis, particularly in complex cases.
However, the research was based on historical medical cases evaluated in a simulated environment rather than real-world clinical practice.
The study did not involve a direct comparison with physicians treating patients in real time, meaning the findings do not establish that the AI system would outperform doctors in actual clinical settings.
