The Rise of AI-Powered Health Prediction, Featuring Epic’s Curiosity
March 6, 2026
Nature Biotechnology examines how AI trained on electronic health records is opening the door to a new era of preventive medicine.
Curiosity—a family of new generative AI models trained on deidentified records from more than 300 million patients in Cosmos—is designed to simulate future patient health outcomes to support use cases from personalized care to hospital resource planning.
Epic’s Rahul Shah describes the potential to move beyond abstract clinical decisions and toward deeply personalized care for each patient. Shah also highlights Epic’s approach to data privacy, noting that Curiosity contains no identifiable personal information and that researchers can query the model without being able to extract patient data. The article points to Cosmos as a unique large-scale resource and to the trust-based governance framework—led by contributing healthcare organizations—that makes it possible.
Read the full article in Nature Biotechnologyhere.
More Epic in the News
Fewer Care Delays, More Cost Savings: Nationwide Diagnostic Image Exchange Is Possible With TEFCA
October 27, 2026
Epic’s Director of Interoperability Matt Doyle explains how nationwide diagnostic image exchange through TEFCA could reduce duplicate scans, lower costs, and speed up care.
Trust Collaboration Across Patient Record System Saves £4.1 Million in 3 Years
October 26, 2026
The collaboration also freed up an estimated 275,000 hours of nursing and phlebotomy time across the two trusts, and spared patients extra blood draws.
