ChatGPT Health now links with Epic to let clinicians pull in patient data
OpenAI is integrating ChatGPT Health with Epic's electronic health record system to let clinicians import patient data and query it with AI. Access is read-only, and the AI cannot write anything back into records.
OpenAI announced that it is connecting ChatGPT Health with Epic, an electronic health record (EHR) system holding data on more than 325 million patients. In some systems, ChatGPT will be built directly into EHR workflows, letting clinicians use it without leaving a patient's chart.
With the integration, clinicians can pull together information such as appointment notes, lab results, medications and specialist documentation. The tool also helps track a patient's history, spot changes over time, and prepare for upcoming appointments. In certain deployments, clinicians can use ChatGPT directly for pre-visit review or to build a clinical timeline.
OpenAI stressed that the integration only grants read-only access to patient records — the AI does not write anything back into the system.
New plugin for public health data
The company is also rolling out a new Healthcare Public Data plugin that draws information from sources including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed and PubMed. It's meant to help healthcare workers pull together details like clinical trial eligibility criteria, medication identifiers and insurance coverage policies.
Organizations that have signed a Business Associate Agreement can now also use ChatGPT Work, Codex, apps and connectors within compliant workflows.
Last month, OpenAI made its ChatGPT health features available to all U.S. consumers, saying at the time that users send roughly 300 million health-related queries to the chatbot every week. Despite the deepening ties with medical systems, OpenAI continues to maintain that its AI is not meant for diagnosis or treatment.
The company said that in testing across 27 clinical use cases with more than 4,300 physician-reviewed responses, 99.1% were rated safe. Still, even a small number of unsafe answers can cause real harm — a Florida pastor recently sued OpenAI, alleging ChatGPT gave him a near-fatal recommendation, and in May the company was also sued by relatives of a user who blamed ChatGPT for faulty dosage advice.
