Microsoft AI Panel Outpaces Doctors in Complex Medical Diagnostics

Microsoft has unveiled a diagnostic system designed to mimic a panel of physicians, claiming it can outperform human doctors on complex medical cases. The technology, known as MAI-DxO or DxO, uses a coordinated network of artificial intelligence agents to question patients, order tests, and debate potential diagnoses. Mustafa Suleyman, the CEO of Microsoft AI, describes this development as a critical step toward achieving 'medical superintelligence,' a state where AI surpasses the reasoning capabilities of current human-led health systems.

The performance data, published in the journal Science this week, highlights a significant gap between the AI system and individual healthcare providers. In a simulation using complex cases from the New England Journal of Medicine, the AI system arrived at the correct diagnosis 83% of the time. In comparison, primary care doctors and general internists, working alone with access to reference materials, were correct only 36% of the time. It is important to note that these specific cases are typically handled by specialists or teams of physicians, who were not included in this particular comparison.

Dominic King, Vice President of Health at Microsoft AI, emphasized the urgency of this technological shift. He noted that the primary challenge in modern healthcare is the widening gap between the demand for medical expertise and the available supply. By publishing these results, Microsoft joins other major tech players like Google and OpenAI in asserting that their AI tools can match or exceed human performance in clinical settings.

The study’s data suggests that much of the performance improvement stems from the underlying power of the AI models rather than just the orchestration system. Microsoft tested the DxO framework with models from various companies, including OpenAI, Google, Anthropic, Meta, and DeepSeek. The highest accuracy scores were achieved when the system utilized OpenAI’s GPT-5 and GPT-5.5 models. While the orchestration layer provided additional benefits, the base model's capability remained the primary driver of results.

Cost analysis reveals a trade-off between accuracy and expense. When using GPT-5 alone with a basic prompt, the system scored 78.9% on complex cases at an estimated cost of $8,400 per case. Adding the DxO orchestration reduced that cost by approximately $3,000 but improved accuracy by only about four points. For emergency department cases, the system showed little difference in either cost or accuracy whether it was assisted by DxO or running independently.

Newer models demonstrate more distinct benefits from the orchestration framework. With GPT-5.5, released in April 2026, the DxO system boosted accuracy on complex cases from 72% to nearly 87% at a slightly lower cost. However, these gains did not translate to emergency department scenarios, where both versions scored around 85% at similar costs. The system proved most effective when paired with less powerful models; for instance, Gemini 2.5 Flash improved from 46.4% accuracy alone to nearly 69% with the assistance of DxO.

Microsoft researchers indicate that accuracy improvements naturally diminish as models approach the theoretical maximum of performance. The company is currently collaborating with the Mayo Clinic to train a dedicated healthcare foundation model, with a release scheduled for next year. Suleyman believes this infrastructure will allow for significant leaps in medical reasoning and is working toward a 'coding style breakthrough' in medical records by 2027, enabling accurate predictions of how patient records might evolve.

To validate these findings in real-world settings, Microsoft has partnered with Beth Israel Deaconess Medical Center in Boston. The upcoming study will evaluate the clinical reasoning capabilities of DxO across diverse hospital environments, including emergency and urgent care settings. Patients will receive standard care while consenting to parallel engagement with the AI system. This approach aims to generate compelling evidence for the large-scale deployment of these diagnostic tools, a move Suleyman believes will differentiate Microsoft from other frontier AI labs in the coming years.