Microsoft’s AI Diagnostic Tool Outperforms Doctors in Complex Medical Cases

Microsoft researchers unveiled new findings on October 22 showing that an artificial intelligence system diagnosed complex medical conditions more accurately than general physicians. The study, published in Science, found that Microsoft’s Diagnostic Orchestrator achieved 83.2% accuracy across challenging medical cases. Physicians using reference materials achieved 36.1%. The technology follows a step-by-step diagnostic process similar to how doctors evaluate patients.

Unlike conventional medical AI systems that analyze patient information simultaneously, Microsoft’s tool asks questions, recommends examinations, and selects diagnostic tests sequentially. It uses a simulated panel of clinical experts to evaluate findings and update potential diagnoses. Researchers tested the system against more than 300 complex medical cases and additional emergency department scenarios. The AI also demonstrated lower estimated diagnostic testing costs.

Microsoft plans to evaluate the technology at Beth Israel Deaconess Medical Center in Boston, focusing on emergency department patient triage. However, researchers acknowledge that controlled testing cannot establish real-world clinical safety or effectiveness. The findings could strengthen Microsoft’s position in healthcare AI as technology companies expand into clinical decision-support systems. Further validation will determine whether the tool can improve hospital efficiency and patient care.