OpenAI recently announced that its ChatGPT for Healthcare, specifically designed for the healthcare industry, has officially supported integration with mainstream Epic electronic health record (EHR) systems. This major advancement means that medical staff can not only access patients' complete medical records within authorized limits, but also use AI to efficiently summarize patient visits, thereby comprehensively optimizing daily clinical workflows.

Looking back at OpenAI's continuous efforts in the healthcare sector, the company has previously launched several targeted products. These include ChatGPT Health for individual users, offering a private space to discuss health conditions and medical issues; followed by ChatGPT for Healthcare based on the GPT-5 model, compliant with the Health Insurance Portability and Accountability Act (HIPAA) standards; and then launched ChatGPT for Clinicians, specifically designed for doctors and other healthcare professionals, aiming to assist with complex daily clinical tasks.

image.png

Through this deep integration with the Epic electronic health record system, ChatGPT breaks through medical data silos, allowing AI capabilities to serve hospitals and medical institutions more deeply. With this integration function, clinical doctors can systematically view patients' appointment records, laboratory test results, medication information, and specialist documents. On this basis, the system can intelligently summarize changes in patients' conditions since their last visit, highlight recent key test results, accurately identify medication adjustments, and proactively organize unresolved follow-up or treatment matters, greatly reducing the administrative burden on doctors.

To flexibly adapt to different hospital information architecture, OpenAI provides two optional deployment methods. The first is to import patient information from the supported electronic health record system into ChatGPT for analysis; the second is to directly embed the AI assistant into existing electronic health record workflows, allowing clinical doctors to seamlessly call the AI assistant without leaving the patient's medical record page.

In addition to the interoperability of medical records, OpenAI also launched a new Healthcare Public Data plugin. This plugin connects ChatGPT to nine official authoritative medical data sources, including PubMed, ClinicalTrials.gov, CMS Coverage, RxNorm, and DailyMed. Medical teams no longer need to log in separately to various external service platforms; they can directly use ChatGPT to efficiently search and compare eligibility requirements for clinical trials, drug instructions, drug identifiers, and insurance coverage policies.

In terms of security and accuracy evaluation, OpenAI invited professional doctors to conduct rigorous testing on 27 real clinical scenarios before launching, including drug review, pre-visit summary, clinical timeline organization, and patient handover. Among 4,363 scores, 99.1% of the answers were rated as safe. In addition, in another evaluation conducted on a large U.S. medical dataset, ChatGPT's responses to the tested data sources also performed well, with over 93% receiving good or higher accuracy ratings.

Currently, eligible ChatGPT for Clinicians users in the United States can install the Healthcare Public Data plugin first. However, it should be noted that this highly anticipated Epic electronic health record integration feature is currently available for institutional-level medical organizations, and individual ChatGPT accounts are temporarily unable to use it.