The Rise of AI-Powered Health Prediction, Featuring Epic’s Curiosity

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 Biotechnology here.