Microsoft has developed advanced artificial intelligence that makes diagnoses like a real doctor

The system is designed to independently guide the individual steps of the diagnostic process and adapt to new information. The artificial intelligence first selects relevant questions about the patient's medical history and determines key elements of the physical examination. Based on the information collected, it then decides whether to order additional diagnostic tests, and uses the results to verify and update its diagnostic hypotheses. This approach allows for a gradual narrowing of the range of possible causes for the patient's problems.

In tests covering hundreds of complex and unusual medical cases, the tool proved to be highly accurate. It outperformed general practitioners who relied on literature and online sources to make diagnoses. The system also performed better when analyzing cases from a large database of emergency room visits. The results suggest that such artificial intelligence could help doctors deal with complex cases and make decisions easier, especially when there is a lot of information to consider.

Despite the encouraging results, experts warn that caution is needed, as the technology needs to be thoroughly tested in everyday clinical settings before it can be widely used. Laboratory results do not necessarily guarantee the same success in working with real patients, where time constraints, incomplete data and unpredictable circumstances are also important. In collaboration with a hospital in Boston, they plan to test the system in emergency department waiting rooms, where they will be able to assess its usefulness in real-world situations.