OpenAI Expands ChatGPT for Healthcare With Epic EHR, Public Health Data Integrations

The updates are designed to let authorized clinicians and healthcare teams access, summarize, and verify clinical and public health information within ChatGPT.

OpenAI has expanded ChatGPT for Healthcare with an integration for Epic’s electronic health record (EHR) and a new plugin connecting the platform to nine official public healthcare data sources. The updates are designed to let authorized clinicians and healthcare teams access, summarize, and verify clinical and public health information within ChatGPT.

Announced Tuesday, the Epic integration allows clinicians to bring authorized patient information into ChatGPT or access ChatGPT within supported Epic workflows. The integration can combine clinical notes, laboratory results, medications and specialist documentation, OpenAI said in a blog post.

The Epic integration is read-only and does not write information back to a patient’s medical record. It supports two workflows: clinicians can pull patient data into ChatGPT to review medical histories, identify changes and prepare for appointments, or use ChatGPT within supported EHR workflows without leaving the patient chart.

UCSF Health is serving as a pilot partner. Suresh Gunasekaran, president and CEO of UCSF Health, said the organization is evaluating how the integration can help clinical teams synthesize complex patient records and identify relevant changes more quickly.

OpenAI has also introduced a healthcare public data plugin that connects ChatGPT to nine official sources, including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed and PubMed. The integration allows teams to work with specific records, fields, identifiers and versions from these sources.

For example, research teams can use ClinicalTrials.gov to identify recruiting clinical trials and compare eligibility criteria, while pharmacy teams can use DailyMed to verify medication labels and warnings.

OpenAI said physicians across 60 countries and 26 specialties have reviewed more than 700,000 model responses to help evaluate its healthcare capabilities. In testing across 27 clinical use cases, physicians rated 99.1% of 4,363 responses as safe. A separate evaluation found that more than 93% of responses for each of five connected data sources were rated as having “good” or better accuracy.

ChatGPT for Healthcare also includes enterprise controls such as role-based access, single sign-on and audit logs. OpenAI said customers with an applicable business associate agreement can use ChatGPT Work, Codex, apps and plugins within the same workspace for HIPAA-compliant workflows.

Stay tuned for more such updates on Digital Health News