A new artificial intelligence tool developed by Microsoft has proven to be more than twice as accurate as general practitioners in resolving complex medical cases. Research results published in the journal Science on October 22, 2026, indicate that the model uses a sequential approach, simulating human clinical reasoning to diagnose patients.
Unlike traditional models that analyze all patient data simultaneously, the new technology employs an "orchestrator." This system coordinates the stages of diagnosis, deciding what questions to ask about the patient's history, which tests to order, and how to adjust working hypotheses as results are obtained. The process involves the collaboration of a simulated panel of clinical experts.
In tests conducted with over 300 unusual medical dilemmas, the AI outperformed general practitioners who used textbooks and the internet as support. The tool also showed superior performance in scenarios based on emergency room admission data. According to Harsha Nori, head of AI at Microsoft AI Health, the study evaluated not only accuracy, but also rigor and care in the diagnostic process, which proved to be less costly than when conducted exclusively by humans.
Dr. Dominic King, corporate vice president of the Microsoft AI health unit, emphasized that the technology seeks to bring AI performance closer to clinical reality, overcoming the limitations of multiple-choice tests. The research team, in collaboration with the Beth Israel Deaconess Medical Center in Boston, now plans to test the tool's capability in patient triage in emergency department waiting rooms.