Microsoft's AI Diagnoses Patients More Accurately Than Doctors

A team of researchers at Microsoft, the American tech company, built an AI-based diagnostic tool. They say it outperforms human doctors.

The study, published in the journal Science, describes how the AI system, named MAI-DxO, works like a team of doctors. Each member proposes a possible diagnosis and critiques the others' work.

The team tested it on hard cases from the prestigious New England Journal of Medicine. The AI got the diagnosis right 83% of the time. Human doctors with access to reference materials got it right 36% of the time. Those without such access got it right only 22% of the time.

In more routine situations, the kind doctors see in hospital emergency rooms, the AI made the correct call 80% of the time. Humans managed only 47%. On top of that, the AI cost several times less.

This multi-agent AI setup, where one agent checks another's work and catches its mistakes, is already used elsewhere too. In coding, for example, one AI reviews code written by another and catches errors the first one missed.

But keep this in mind: this is research work based on controlled experiments. Clinical trials are only just starting, so far at a single hospital in Boston. So do not expect this AI to show up at your local clinic anytime soon. The numbers claiming it beats human doctors deserve a cautious read.