Mayo Clinic’s AI chief sees its partnership with Microsoft as a chance to do something the health system’s growing collection of AI point solutions cannot: learn from the whole patient.
Rochester, Minn.-based Mayo and Microsoft announced June 2 that they are collaborating to develop a frontier AI model purpose-built for healthcare. The model is designed to bring together different types of clinical data to support earlier diagnoses, more personalized treatment decisions and improved patient outcomes.
Micky Tripathi, PhD, chief AI implementation officer at Mayo Clinic, told Becker’s the work could ultimately bring together insights that today are often generated by separate AI models focused on individual diseases, organs or specialties.
“If you think about all the solutions I was just talking about … they’re all kind of point solutions targeting a particular disease condition, a particular specialty, a different part of you, without looking at the whole of you,” Dr. Tripathi said.
Mayo has 128 clinical AI solutions in practice, including tools focused on areas such as cardiovascular risk, dementia and pancreatic cancer detection.
The Microsoft partnership takes a different approach.
“What if we looked at every patient and their entire story?” Dr. Tripathi said.
Rather than focusing only on patients who received a particular test or developed a particular condition, he said the concept is to learn from patients’ longitudinal journeys and the different types of information generated along the way.
That could include imaging, laboratory results, electrocardiograms, electroencephalograms, genetic information and clinical events such as emergency department and primary care visits.
Dr. Tripathi compared the concept to the way a foundation model learns patterns from large amounts of language. In this case, however, the model would learn patterns across medical events and different forms of clinical data.
A clinician could eventually provide information about an individual patient, such as their age, symptoms and health history, and use the model to assess different possible scenarios based on patterns learned from other patients, he said.
That could include assessing the probability of different diagnoses. Once a condition is identified, Dr. Tripathi said the model could potentially help clinicians explore how patients with similar characteristics might fare under different treatment approaches.
He described a scenario in which a physician could discuss several treatment options with a patient and consider what the data suggests about the potential trajectory associated with each one, alongside factors such as quality of life and treatment preferences.
The approach could also offer another way to examine rare or unusual cases. Dr. Tripathi said a model drawing from years of longitudinal information could potentially surface a small number of previous patients whose circumstances closely resembled the patient being treated and help clinicians understand what happened in those cases.
“What have we learned from all the patients that allow us to look at you as an individual and say, here’s what we know about you now?” he said.
Dr. Tripathi described the idea as “putting the precision into precision medicine.”
The partnership combines Mayo’s clinical expertise, deidentified clinical health data and longitudinal insights with Microsoft’s AI, cloud and engineering capabilities. Mayo will retain ownership of the model, according to Microsoft’s June announcement.
The model is initially being deployed within Mayo’s clinical environment, where the organizations plan to test and refine it through real-world use. Microsoft also plans to make the model available through Azure Foundry APIs, which could eventually give other healthcare organizations access to the technology.
For Dr. Tripathi, the work represents an opportunity to connect areas of AI that are largely being developed separately today.
“It’s kind of bringing it all together,” he said. “How do I think about this holistically so that I can help treat the whole person?”
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
