Epic unveils AI model for clinical foresight

Epic announced a new AI tool designed to help clinicians anticipate patient needs and improve decision-making.

Here are six things to know about the tool:

The platform, called Comet, uses generative models trained on more than 100 billion medical events from Epic’s Cosmos database, according to a Sept. 3 news release. By simulating possible patient outcomes, Comet aims to give care teams data-driven insights into what may happen next during a patient’s health journey.
Epic said the technology could assist with common decisions such as whether to admit a patient with abdominal pain, determine the safety of early discharge or estimate a patient’s risk for readmission.
Comet generates multiple possible scenarios and summarizes them into insights for clinicians within their workflows.
The models are trained on de-identified data from millions of patient records. Epic said Comet is built within Cosmos, which operates under strict privacy and compliance standards. Because Cosmos is continuously updated, the company noted, the data reflects evolving clinical practices.
In early testing across 78 evaluations, Comet performed as well or better than most single-task predictive models, according to the release. Beginning in February 2026, researchers at health systems participating in Cosmos will be able to explore the tool in a virtual lab.
Epic said Comet’s simulations could be used to forecast risks such as extended hospitalization, 30-day readmission, cardiovascular disease, or the onset of conditions including pancreatic cancer. The company added that the model is expected to improve as it incorporates more data types.