OpenAI connects Epic records and nine public healthcare datasets to ChatGPT for Healthcare
OpenAI said Tuesday that healthcare organizations can now bring authorized patient context from Epic into ChatGPT for Healthcare, and can query nine official public healthcare datasets through a new plugin, the company wrote.
A clinician can ask what changed since last visit, which recent labs to review, or which medications moved. ChatGPT summarizes and points back to the supporting chart information. In supported deployments it also sits inside the EHR layout, so the clinician does not leave the patient chart.
The AI can read authorized chart context. It is not supposed to write anything back into Epic. An electronic health record is the hospital’s file on the patient.
UCSF Health is the pilot partner. Suresh Gunasekaran, the health system’s president and CEO, said on OpenAI’s post that the integration is meant to cut time spent synthesizing data and leave more time with patients.
The Healthcare Public Data plugin covers nine official sources. OpenAI names ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed among them. Teams work with specific records, fields, identifiers, and versions.
Ivan Mehta at TechCrunch writes that Epic holds data for over 325 million patients. That figure is TechCrunch. The OpenAI post does not print it.
I keep 325 million on TechCrunch.
TechCrunch also says ChatGPT handles 300 million health-related queries a week, a company figure via that story, after last month’s consumer health rollout to U.S. users. That consumer rollout is a different product.
Heather Landi at Fierce Healthcare notes ChatGPT for Healthcare launched in January and ChatGPT for Clinicians in April. The product is ChatGPT for Healthcare, not a new ChatGPT Health consumer ship.
OpenAI says it partnered with hundreds of physicians across 60 countries, 49 languages, and 26 medical specialties, and reviewed more than 700,000 model responses. On EHR context, 4,363 ratings across 27 clinical use cases, with physicians rating 99.1 percent of responses safe.
A separate two-round eval on five connected data sources put more than 93 percent of responses at good or better accuracy. Those are OpenAI’s ratings, not a regulator’s.
ChatGPT for Healthcare customers can have an admin enable the EHR integration and the Public Data plugin. Eligible U.S. ChatGPT for Clinicians users can install Healthcare Public Data. The EHR integration is not available for individual accounts.
With a Business Associate Agreement, customers can use ChatGPT Work, Codex, apps, and plugins for HIPAA-compliant workflows, OpenAI says. Role-based access, SSO, and audit logs come with the workspace.
UCSF is the frontline check, not a rollout total.
Pingback: Nexus put more than 153 million drivers-license scans up for sale. ChatGPT for Healthcare can read Epic charts, not write them.