Microsoft has introduced the Microsoft AI Diagnostic Orchestrator (MAI-DxO), a medical AI system that outperformed human doctors in diagnosing complex cases published in the New England Journal of Medicine. In a study shared on arXiv, MAI-DxO achieved an 85% diagnostic accuracy compared to 20% by doctors. The system analyzed about 300 challenging cases, using a stepwise process that allowed it to simulate real-world clinical reasoning, including test ordering and decision-making at each stage. Microsoft also measured cost-effectiveness, and MAI-DxO was able to make accurate diagnoses at 20% lower cost on average.
Mustafa Suleyman, CEO of Microsoft AI, stated, “This is a startling result. I think it gives us a clear line of sight to making the very best expert diagnostics available to everybody in the world at an unbelievably affordable price point.” Microsoft built MAI-DxO by combining outputs from existing AI models and incorporating a virtual consultation tool to reflect how doctors collaborate. The system explains its diagnostic process, offering transparency for clinicians. Dr. Eric Topol, Director and Founder of the Scripps Research Translational Institute, commented, “The four-fold increase in accuracy was more than previous studies have shown. Most of the time there is a 10% absolute percentage difference, so this is a really big jump.” While the technology is still in development, experts say real-world trials and regulatory clarity are needed before clinical use.
