Dr. Bimal Desai of CHOP walks through the three waves of Epic AI tools, how CHOP partners on validation and how it stages every AI rollout. Watch below or on YouTube.
In a LinkedIn post after Epic’s user group meeting in August, Dr. Bimal Desai, Vice President and Chief Health Informatics Officer at Children’s Hospital of Philadelphia (CHOP), wrote that the more than 20 Epic generative AI tools CHOP had implemented, with 10 more in validation, felt a bit disconnected. On a recent episode of the healthsystemCIO Show, Desai said the next stage is an assistant that anticipates what a clinician needs during a visit. He also described how CHOP validates new features with Epic, when it buys outside the EHR suite and how it introduces each new tool to clinicians.
The disconnect traces back to the order in which the tools arrived, Desai said. When large language models took off in late 2022, Epic started with the lowest-risk uses. Specifically, those included automated draft replies to patient portal messages and a text assistant that rewrites selected text. The second wave then went after time-consuming work, such as ambient scribes and summaries of long outpatient records. That order was no surprise, Desai said, and he did not offer the disconnect as a criticism of the EHR.
The third wave is an assistant that follows the whole visit. Desai described a visit in which a physician starts asking about rashes, fevers and joint swelling. A well-built assistant would recognize that line of questioning and surface the patient’s latest white count, inflammatory markers and family history, along with how similar patients in Epic’s Cosmos data presented.
Source link







