Microsoft’s AI Diagnostic Panel Beats Doctors in Complex Case Simulation

Microsoft has unveiled a new AI diagnostic system that correctly identified complex medical conditions in a simulation at more than double the rate of human generalist physicians. The system, known as MAI-DxO, operates as a coordinated panel of artificial intelligence agents that mimic the workflow of a medical team. These agents question patients, order necessary tests, and debate potential diagnoses before arriving at a conclusion. Mustafa Suleyman, the CEO of Microsoft AI, described the achievement as a critical step toward achieving what the company calls “medical superintelligence.”

The performance data, published in the journal Science on Thursday, shows a significant gap between the AI and human performance on difficult cases. When tested against complex cases from the New England Journal of Medicine, the AI system was accurate 83 percent of the time. In comparison, primary care doctors and general internists working individually with reference materials were correct only 36 percent of the time. It is important to note that these specific cases are typically handled by specialists or teams in real-world settings, who were not included in the direct comparison.

Dominic King, Vice President of Health at Microsoft AI, highlighted the urgency of this development. He noted that the current healthcare system faces a growing mismatch between the demand for medical expertise and the available supply. By publishing these results, Microsoft joins Google and OpenAI in asserting that their AI tools can match or exceed human diagnostic capabilities. The study’s data indicates that much of the improvement stems from more powerful underlying AI models, with the orchestration system providing additional gains on top of the base performance.

Microsoft evaluated the system using models from several major technology companies, including OpenAI, Google, Anthropic, Meta, and DeepSeek. The most favorable results were achieved using OpenAI’s GPT-5 and GPT-5.5 models. The company also measured the cost of running the AI for each case. While GPT-5 alone scored 78.9 percent on complex cases at an estimated cost of $8,400, the addition of the diagnostic system reduced the cost by approximately $3,000 while raising accuracy by only four points. On emergency department cases, the improvement in both accuracy and cost was minimal.

The effectiveness of the system varies depending on the underlying model and the complexity of the case. With the newer GPT-5.5 model, released in April 2026, the diagnostic system boosted accuracy on complex cases from 72 percent to nearly 87 percent at a slightly lower cost. However, these gains did not translate to emergency department cases, where both versions scored around 85 percent. The system proved most beneficial when paired with less capable models, such as Gemini 2.5 Flash, which improved from 46.4 percent to nearly 69 percent. Researchers noted that accuracy gains naturally diminish as models approach their theoretical maximum.

Suleyman, who previously co-founded DeepMind before joining Microsoft in 2024, is leading the company’s effort to build its own frontier AI. He stated that Microsoft is collaborating with the Mayo Clinic to train a healthcare-specific foundation model, with a release planned for next year. He expects this model to enable significant advancements, allowing the company to stream highly accurate predictions about how medical records might unfold. Suleyman believes this foresight will be invaluable for clinicians and anticipates an order of magnitude improvement in healthcare model capability within the next 24 to 36 months.

To validate the technology in real-world settings, Microsoft has established a partnership with Beth Israel Deaconess Medical Center in Boston. The goal is to evaluate the clinical reasoning capabilities of the system across various hospital environments. Patients will receive standard treatment, including triage and specialist consultations, while consenting patients will also interact with the AI. The company will then compare the AI’s recommendations against those of human doctors. King described this as potentially the largest study of its kind, aiming to provide compelling evidence for the future deployment of these systems at scale.

The study coincides with a broader manifesto from Microsoft defining “medical superintelligence” as the ability to surpass the strongest contemporary human-led health systems in reasoning and coordination. The company aims to move beyond benchmarking individual tasks and instead integrate with clinicians for better overall outcomes. Microsoft already sells dictation and radiology tools and views this new diagnostic system as an expansion of that portfolio. Suleyman emphasized that the company is committed to having the leading medical clinical decision support products in the market over the next few years to accelerate the introduction of this technology to the world.