OpenAI announced today the integration of ChatGPT Health with Epic’s electronic health record (EHR) system, which contains data for over 325 million patients. This integration will enable clinicians to import patient data and utilize AI for queries. The company also stated that in certain systems, ChatGPT will be directly embedded into existing EHR workflows.
Through ChatGPT, clinicians can now effortlessly access and summarize various types of information, including appointment notes, lab results, medication details, and specialist documentation. They can also review patient histories, identify changes over time, and prepare more effectively for upcoming appointments. OpenAI indicated that in some implementations, clinicians can access ChatGPT directly within the patient chart for pre-visit reviews and to construct clinical timelines, without having to leave their current view. The company emphasized that this integration provides only read-only access to health records, meaning the AI cannot write or alter any data.
Additionally, OpenAI is introducing a new Healthcare Public Data plug-in, designed to retrieve information from sources such as ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. This new plug-in aims to assist healthcare professionals in synthesizing data related to trial eligibility criteria, medication identifiers, coverage policy versions, and provider records.
Beyond these new functionalities, the company is enabling organizations with a Business Associate Agreement to utilize ChatGPT Work, Codex, applications, and connectors within their workspaces for compliant operations.
Last month, OpenAI made ChatGPT for health available to all U.S. consumers. At that time, the company reported that users were submitting 300 million health-related queries to ChatGPT weekly. Despite deep integrations with medical systems, OpenAI consistently asserts that AI should not be used for diagnosis or treatment.
OpenAI conducted a study collecting over 4,300 responses from physicians across 27 clinical use cases, including pre-visit review, clinical timelines, medication review, and handoff summaries, finding that 99.1% of these responses were deemed safe. However, the company acknowledges that even a small number of unsafe answers could potentially lead to adverse outcomes for patients.
Prior to this rollout, OpenAI faced legal action from a Florida-based pastor who claimed that ChatGPT provided him with a nearly fatal recommendation. In May, the company was also sued by family members of a user who attributed wrongful dosage advice to ChatGPT.