17 Apr Epic, Microsoft bring GPT-4 to EHRs
HIMSS 2023 kicked off Monday in Chicago. Follow our live blog for the latest updates and happenings from the conference.
Epic Systems is working with Microsoft to integrate generative AI technology into its electronic health record software for the first time, the companies said Monday.
The announcement was made in conjunction with the first day of the HIMSS conference, which is being held in Chicago this week.
Health systems using Epic’s EHR system will be able to run generative AI solutions through Microsoft’s OpenAI Azure Service. Microsoft uses OpenAI’s language model GPT-4 capabilities in its Azure cloud solution.
Related: Unpacking ChatGPT’s early uses in healthcare
While Epic has worked with machine learning in the past, it’s the first time it has used generative AI, said Seth Hain, senior vice president of research and development.
“There is a real potential when integrating it into workflows to increase the productivity of physicians in the exam room, folks in the back office [and] schedulers up front,” Hain said
Epic and Microsoft said there were two initial AI-enabled solutions developed in the initial roll out of the integration. The first use case occurs through Epic’s In Basket communication software system. Through the integration with Microsoft, clinicians can use generative AI to create a draft response when communicating asynchronously with patients.
The first customers for this use case are live on the platform now. Madison, Wisconsin-based UW Health and UC San Diego Health have already rolled out the company’s AI for In-Basket responses to a limited number of users. Stanford Health Care in California is expected to add the functionality soon.
Hain said the company will roll out the software to general availability in “weeks and months” but it is not committing to a firm launch date. It expects at least one additional round of test customers.
The second use case allows providers to use generative AI to source recommendations from Epic’s Slicer Dicer data visualization tool. Typically, users have to customize specific data searches on their own. With the generative AI capabilities, users can type something and the system will automatically recommend different metrics. A spokesperson said this functionality is still in development. It is expected to launch to an initial group of end users later this year.
According to Hain, Epic deployed “a couple dozen” engineers to incorporate generative AI into the company’s software. The five to six month project tapped engineers in other departments once core functionality was built out. Microsoft handled the core functionality of GPT-4 algorithms, he said.
Not a Modern Healthcare subscriber? Sign up today.
The move by Epic comes at a time when generative AI is having its moment in the sun. A rush of digital health companies are seeking to cash in on the popularity of OpenAI’s ChatGPT tool, which is free for all people to use and has amassed more than 100 million users.
But experts are unsure how ChatGPT and GPT-4 will influence clinical diagnosis and decision making. Medical AI experts say the first wave of adoption will take place in areas where there are administrative redundancies. This follows Epic’s line of thinking, Hain said.
“It’s key that [generative AI] is done in a responsible way, with humans in the loop,” Hain said. “I think that is a generally true statement, but it is critical in the context of healthcare. We’re making sure that’s the approach in these contexts.”
Last month, Nuance Communications, a clinical documentation software company owned by Microsoft, added OpenAI’s ChatGPT successor GPT-4 to its latest voice transcription application. The software, according to the company, can summarize and enter conversations between clinicians and patients directly into EHR systems using OpenAI’s GPT-4 generative AI capabilities.
Hain said it was too early to determine the role of third-party software leveraging generative AI in the EHR.
“We already see organizations starting to [interoperability standards] those to build these types of integrations and applications,” Hain said, “I expect there will be—as there always is—an evolution of those standards and techniques, as more and more people acclimate to generative AI.”
This story first appeared in Digital Health Business & Technology.