Clinicians using ChatGPT for Healthcare can now bring authorized Epic patient records into the AI workspace without manually moving information between systems.
OpenAI announced Tuesday that supported Epic electronic health record environments can connect with ChatGPT for Healthcare, giving authorized users access to information such as appointment notes, lab results, medications and specialist documentation. The integration can also appear within supported EHR layouts for tasks such as pre-visit reviews and clinical timelines.
In supported deployments, ChatGPT can also appear directly inside an EHR layout. That could let clinicians perform tasks such as pre-visit reviews and build clinical timelines without switching between applications. Importantly, the connection is read-only.
ChatGPT can retrieve and summarize authorized information, but it does not write AI-generated content back into the patient record.
UCSF Health is serving as a pilot partner. “As a pilot partner, we’re exploring how the new EHR integration with ChatGPT for Healthcare can help clinical teams understand what has changed and what matters most across a complex patient record,” UCSF Health President and CEO Suresh Gunasekaran said in OpenAI’s announcement.
A broader medical data layer
OpenAI is also introducing a Healthcare Public Data plugin that connects ChatGPT to nine official healthcare data sources.
The sources include ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed and PubMed. The company says teams can use the connections to examine details such as clinical trial eligibility, drug identifiers, coverage policies and provider records.
That expands ChatGPT beyond patient-specific information, giving researchers, pharmacy teams and population health groups another way to combine authoritative medical data within the same workspace.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
Safety remains the sticking point
OpenAI said physicians evaluated ChatGPT across 27 clinical use cases involving connected EHR information. Across 4,363 ratings, 99.1% of responses were rated safe.
The company also reported accuracy rating responses above 93% across each of the five public data sources tested, with DailyMed scoring 98.6% and CMS Coverage 93.2%.
Those numbers are encouraging, but they do not eliminate the risk of an incorrect answer making its way into clinical decision-making. Recent lawsuits alleging harmful medical advice from ChatGPT also underscore the consequences when AI gets health-related information wrong.
Why the integration could matter
Clinicians already spend significant time gathering facts from different parts of a patient’s chart. An AI system that can rapidly surface changes and supporting evidence could reduce that administrative burden. But because the system remains read-only, human clinicians still have to verify the information and decide what belongs in the patient’s care.
For hospitals, the integration positions ChatGPT more as a clinical review and decision-support layer than an autonomous medical system. OpenAI said healthcare customers can deploy it within governed workspaces that include role-based access, single sign-on and audit logs.
The read-only design is therefore as important as the AI capability itself. ChatGPT can help clinicians find and summarize information across complex records, but the clinician remains responsible for verifying that information and deciding how it should affect patient care.
Other news: Apple is reportedly exploring screenless wearables, new form factors, and more price tiers as it considers one of the biggest overhauls of its wearable lineup since the Apple Watch debuted.