Microsoft's new neural network revolutionizes medical diagnosis

Microsoft has introduced an innovative neural network for medical diagnosis, surpassing traditional methods in efficiency.

The system operates like a virtual consult: one agent proposes diagnoses, another challenges the version, and the third assesses the need for expensive tests. AI independently asks clarifying questions and adjusts hypotheses as new data is obtained.

The system's accuracy on complex cases from the New England Journal of Medicine was 83% compared to 36% for general practitioners with internet access.

• On complex clinical cases, AI accuracy reached 83%, while doctors achieved 36%
• On typical emergency room cases, the result was 80% versus 47% for doctors
• The neural network is four times more effective at diagnosing diseases than doctors

The development demonstrates a significant breakthrough in the application of artificial intelligence in healthcare, potentially changing primary diagnostic standards.