Mayo Clinic and Cerebras Systems will develop and train large language models to determine best treatments for individual patients with specific conditions.
Mayo Clinic has teamed with Cerebras Systems to use its generative AI technology to develop and train large language models that can be deployed to better diagnose and improve outcomes for diverse conditions and act as the first patient-centric healthcare AI solutions for specific medical applications.
The companies unveiled their multiyear collaboration at the JP Morgan Healthcare Conference on January 15 in San Francisco, saying that they have already started work on LLMs using Mayo Clinic’s repository of decades-long longitudinal data and Cerebras’ CS-2 system, which is powered by the world’s largest and fastest AI processor, to train them across various languages and specific domains.
The two are starting with the design of a diagnostic model for rheumatoid arthritis (RA) that will recommend the best treatments and therapies for patients based on a combination of de-identified clinical information in patient records, genomic data from DNA, and drug molecules.
“The state-of-the-art AI models we are developing together will work alongside doctors to help with patient diagnosis, treatment planning, and outcome estimation,” said Andrew Feldman, CEO and co-founder of Cerebras, in a statement.
Mayo and Cerebras will use the RA model as a prototype to develop similar ones for other disease states. These models will assist physicians and be trained on genomic data from more than 100,000 patients.
According to the companies, the high-powered computing of this information could show how specific patients respond to treatments based on their genetic makeup, allowing providers to find the appropriate care regimens for each individual faster.
“How do you make the right decision for each patient?” said Dr. Matthew Callstrom, Mayo Clinic’s medical director for strategy and chair of radiology. “You have to weigh all their individual health factors and have extensive experience making treatment decisions anticipating the outcome of therapeutic options. AI will augment that experience with data from millions of patient records.”
