Mayo Clinic, Microsoft join forces on frontier health AI model

Mayo Clinic and Microsoft are collaborating to develop and deploy a healthcare-specific frontier AI model. The model will be designed to enhance clinical decision-making, including supporting earlier diagnoses and providing personalized treatment options, the companies said.

The collaboration will bring together Mayo Clinic's de-identified clinical data and longitudinal health data insights and Microsoft's advanced AI, cloud and engineering capabilities. According to the organizations, they are developing a frontier AI model for healthcare that can provide "the broadest scope of clinical reasoning and healthcare use cases."

"Frontier medical intelligence is around the corner," said Mustafa Suleyman, CEO of Microsoft AI, in the press release. "This is the best collaboration imaginable to help us accelerate toward that future. Mayo has unparalleled clinical expertise, de-identified clinical health data and longitudinal medical insights, and we're thrilled to partner with their world-class physicians to build a state-of-the-art foundation model for healthcare."

Frontier AI models are highly advanced. According to an Nvidia blog, the models are trained on massive datasets and can perform many tasks. The new model that Microsoft and Mayo Clinic will develop will be trained on diverse clinical data and deployed within the health system's clinical environment for testing and improvement.

Notably, Mayo Clinic will own the frontier AI model, while Microsoft will control the deployment. Microsoft will make the model available worldwide through Azure Foundry APIs.

"Now, by combining our clinical expertise and data foundation with Microsoft's engineering and AI capabilities, we are once again building something new in healthcare and bringing more of Mayo Clinic to more patients," said Gianrico Farrugia, M.D., president and CEO, Mayo Clinic, in the press release.

The announcement comes as questions around AI's clinical decision-making abilities swirl.

While some research shows that AI models struggle with clinical reasoning, particularly in identifying differential diagnoses, a recent study found that an advanced large language model outperformed physicians across various clinical reasoning tasks.

In the latter study, researchers concluded that the OpenAI o1 series "eclipsed most benchmarks of clinical reasoning." However, they also emphasized that these results do not mean that AI can autonomously practice medicine.

Despite these concerns, one example of autonomous medical AI is showing promise in early trial results. In Utah, a pilot is testing the use Doctronic's AI system to autonomously renew prescriptions. Early data shows that physicians agreed with the AI's suggested prescription renewal in 91% of cases that were reviewed.

Anuja Vaidya has covered the healthcare industry since 2012. She currently covers healthcare IT and innovation, including artificial intelligence, digital healthcare, EHRs and interoperability.