Cerebras and Mayo Clinic Unveil Advanced Genomic AI Model 

The intersection of AI and healthcare reached an exciting new milestone this week, as a new and groundbreaking genomic foundation model was unveiled at the JP Morgan Healthcare Conference in San Francisco. 

Cerebras and Mayo Clinic have collaborated to develop a world-class genomic foundation model aimed at harnessing the power of advanced AI and HPC to transform genomics, a field rapidly becoming central to personalized healthcare. The new genomic foundation model is designed to improve diagnostics and personalize treatment selection, with an initial focus on Rheumatoid Arthritis (RA), Cerebras said in a release. 

Treating this condition poses a considerable clinical challenge, frequently involving trial and error to identify the right medications for each patient. Conventional methods that focus on individual genetic markers have struggled to reliably predict how patients will respond to treatment. 

The genomic model was trained with a blend of Mayo Clinic’s comprehensive patient exome data with publicly available human reference genome data, which is a digital DNA sequence representing a composite, "idealized" version of the human genome that serves as a standard framework for human genome comparison to identify genetic variations. 

Cerebras notes this is a contrast to models exclusively trained on the human reference genome and claims its genomic foundation model demonstrates significantly better results on genomic variant classification because it was trained on data sourced from 500 Mayo Clinic patients. The team anticipates further strides in the model’s accuracy as additional patient data is incorporated into the training process. 

The development of the genomic model, which previously would have spanned several years, was significantly expedited by leveraging the Cerebras AI platform for training and customization, the companies said, emphasizing how the Mayo Genomic Foundation Model represents significant steps toward enhancing clinical decision support and advancing precision medicine. 

Dr. Matthew Callstrom, Chair of Radiology in the Midwest and Medical Director for Mayo Clinic's Strategy Department and Generative AI Program, highlighted the transformative potential of the new AI model, noting that it empowers clinicians to make faster, more precise treatment decisions while easing the physical burden on patients. 

“Our clinicians will be able to make more informed decisions based on genomic data, significantly reducing the time it takes to find the right treatment and – more importantly – reducing the physical toll on patients,” he said. “Mayo is committed to fundamentally transforming healthcare by using the most advanced AI technology. We’ve been able to develop AI tools with such promise in less than a year, in part, because of our collaboration with Cerebras that enabled us to create this state-of-the-art AI model for genomics.” 

Along with the new genomic model, the team also designed new benchmarks to evaluate the model's clinically relevant capabilities, such as detecting specific medical conditions from DNA data, addressing a gap in publicly available benchmarks, which focus primarily on identifying structural elements like regulatory or functional regions. 

The companies claim the Mayo Clinic Genomic Foundation Model demonstrates state-of-the-art accuracy in several key areas: 68-100% accuracy in RA benchmarks, 96% accuracy in cancer-predisposing prediction, and 83% accuracy in cardiovascular phenotype prediction, according to the new benchmarks. 

Cerebras Field CTO Natalia Vassilieva said the new model excels in standard tasks like predicting functional and regulatory properties of DNA and also enables discoveries of complex correlations between genetic variants and medical conditions. 

“Unlike current approaches focused on single-variant associations, this model enables the discovery of connections where collections of variants contribute to a particular condition,” she said.