OpenAI has decided that the medical field needs more AI, so it's integrating ChatGPT Health with Epic's electronic health record (EHR) system, which apparently holds data for over 325 million patients. That's a lot of charts to summarize. Now, clinicians can import patient data and ask AI questions, because who needs to read through years of medical history when a chatbot can do it in seconds?

With this new integration, doctors can access appointment notes, laboratory results, medications, and specialist documentation, and have ChatGPT summarize them. It can also look at patient history, identify changes, and prepare for future appointments. And in some deployments, clinicians can even get pre-visit reviews and build clinical timelines without leaving a patient chart. Because why leave the chart when AI can do everything?

For those worried about AI writing back, fear not: the integration only allows read-only access to health records. AI doesn't write anything back. It's like having a really smart intern who only reads and summarizes, never offers opinions or suggestions. Well, at least not in writing.

OpenAI is also adding a Healthcare Public Data plugin, which can fetch info from ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. That's a mouthful, but it means healthcare workers can synthesize data for trial eligibility criteria, medication identifiers, coverage policy versions, and provider records. Because who needs to search multiple databases when a plugin can do it all?

And for organizations with a Business Associate Agreement, they can now use ChatGPT Work, Codex, apps, and connectors for compliant workflows. Because HIPAA compliance is so much easier when you can just click a button.

Last month, OpenAI rolled out ChatGPT for health to all U.S. consumers, and apparently people are asking 300 million health-related queries every week. That's a lot of questions, and presumably a lot of "I'm not a doctor, but..." responses. Despite the deep integration with medical systems, OpenAI maintains that AI is not suitable for diagnosis or treatment. You know, just in case anyone thought otherwise.

OpenAI says it collected over 4,300 responses from physicians across 27 clinical use cases, including pre-visit review, clinical timelines, medication review, and handoff summaries, and found that 99.1% of responses were safe. That sounds impressive, but remember: even a few unsafe answers can lead to harmful results. It's like saying a parachute works 99.1% of the time. Would you jump out of a plane?

And speaking of unsafe, a Florida-based pastor sued the company, alleging that ChatGPT gave him a near-fatal recommendation. Also, in May, family members of a user blamed ChatGPT for giving wrongful advice related to dosage. But hey, 99.1% is safe, right?