Hospital and health system CIOs and chief information security officers have mixed reviews of Epic Systems integrating generative AI technology into its EHR software; while some say it is a win for provider efficiency, others caution that this emerging technology will require massive clinical oversight.
On April 17, Epic said it would extend its partnership with Microsoft to develop and integrate generative AI into its EHR software.
Under the partnership, the companies will combine Microsoft's Azure OpenAI Service with Epic's EHR software with the aim of using generative AI to help healthcare organizations increase productivity and enhance patient care.
J.D. Whitlock, CIO of Dayton (Ohio) Children's Hospital, said the EHR vendor is in a good place to capitalize on this new partnership, as Epic has been using Azure to power its Nebula cognitive computing platform for several years.
"I think I can safely speak for the vast majority of the Epic community that we will be very happy to have, within the next 6 months or so, GPT-4 powered suggested responses to messages in providers' In Baskets," Mr. Whitlock told Becker's.
Mr. Whitlock said the integration will be a huge time-saver for providers and will help them be more efficient in their daily EHR workflows.
"In the industry, there's some uncertainty about ChatGPT attempting to practice medicine, but an In Basket message that is then reviewed by the provider is a safe win for provider efficiency," said Mr. Whitlock.
Tom Barnett, chief information and digital officer of Memphis, Tenn.-based Baptist Memorial Health Care, echoed Mr. Whitlock's sentiments, stating that he is excited for the partnership.
"Tools like ChatGPT and the inevitable clones that will no doubt follow have tremendous potential to be potent force multipliers in most aspects of, not just physicians, but all of our lives," Mr. Barnett told Becker's.
Mr. Barnett said the integration will have the ability to exponentially extend a clinician's reach and productivity.
"The capability to use this type of generative AI to keep close tabs on an entire patient population as well as summarize individual encounter notes all while simultaneously cross examining most academic literature and research studies to cite within visit documentation in near real-time is just the type of major accelerator healthcare could benefit from," said Mr. Barnett.
The first developments of Epic and Microsoft's AI tool are already being piloted by UC San Diego Health, Madison, Wis.-based UW Health, and Palo Alto, Calif.-based Stanford Health Care to draft responses to messages physicians receive from patients through MyChart.
Proceed cautiously and verify accuracy
Generative AI such as ChatGPT does draw concerns from health IT executives when it comes to privacy, security and accuracy.
"The promise of AI does not come without peril," Esmond Kane, chief information security officer of Dallas-based Steward Health Care, told Becker's. "The necessary access to the data supporting these models raises significant privacy concerns, and healthcare is notorious for cumbersome legacy technology ill-prepared for AI."
Mr. Kane said any digital transformation, including any adoption of AI, must be built on solid best practices and with support from regulators to overcome some of its inherent challenges. This is the key to building trust between providers and patients when using these kinds of technologies, according to Mr. Kane.
"These two market leaders must align and be mindful to maintain proper clinical oversight, while ensuring that development operations teams do not over-rely on ChatGPT recommendations," Raymond Lowe, senior vice president and CIO of Commerce, Calif.-based AltaMed, told Becker's. "Using ChatGPT requires continued refinement as organizations will need to safeguard that there are no inherent biases that could inadvertently exacerbate health equity gaps in care."
While generative AI technology can assist healthcare organizations with automation and efficiency, the technology has a tendency to fabricate responses, which makes it a risk to use in clinical settings beyond transcription or automation.
Robert Bart, MD, chief medical information officer at Pittsburgh-based UPMC, told The Wall Street Journal to consider generative AI technologies as "fancy toys" that aren't ready for disease diagnosis just yet, but can be used to improve operational processes such as patient scheduling and workflows.
Currently, most hospitals and health systems are using generative AI technology to do just that — summarizing patient encounters from recorded audio during patient visits to help providers ease documentation burdens.