Health systems across the country are working to determine the best ways to leverage generative AI to alleviate clinical burnout and improve efficiency. UNC Health is an example of one of these systems — it recently announced that it is piloting an in-house generative AI app called Ava, which stands for “AI virtual assistant.”

The app is a conversational bot hosted in UNC’s secure internal environment. When building the app, UNC fed the AI algorithm the health system’s entire training and education library. This allows staff members to “easily and conversationally query the AI tool” instead of searching through vast training libraries, said Dr. David McSwain in a recent interview. He is UNC’s chief medical informatics officer, as well as a pediatric critical care physician at UNC Children’s Hospital.

“Team members can ask questions about the tools that are available to them within Epic and beyond Epic, as well as other health IT tools that are available to our staff. They can ask things like how to use the tool most effectively or how to use a specific functionality. Rather than getting a result back that is 10 different links, they’ll get a concise explanation,” Dr. McSwain explained.

In his view, the tool has significant potential to reduce clinical burnout, given that information overload is a major source of this problem. Health systems hold seemingly endless repositories of information in their electronic health record and training libraries, and UNC’s new app seeks to take that immense amount of data and concisely summarize it. The tool also aims to present that information to clinicians “in a way that’s usable for them at the moment that they need it most,” Dr. McSwain pointed out.

About 30 UNC staff members are participating in the initial pilot for the app. Because the tool is conversational and easy to use, both clinicians and administrators are participating in the trial run. The health system has plans to roll out the app more broadly across its facilities later this year, Dr. McSwain said.

But before the app is rolled out to more staff members, UNC will have discussions with the pilot participants to learn more about their experiences and hear which use cases stood out to them as the most important.

“For me, one of the most exciting things about this tool is how our teammates are going to figure out the best ways to use it for their work. This technology empowers them to pursue their own ideas and to bring them to reality in terms of how they can improve the care of our patients,” Dr. McSwain declared.

Since the use of generative AI is still nascent in the healthcare world, there have been some worries about how providers can deploy the technology safely. Dr. McSwain pointed out that “one of the most important aspects” of UNC’s app is that it is hosted by the health system’s own secure environment. Because of this, the app can handle sensitive information, such as patients’ health data, he said. 

There would be a lot more concerns about data breaches and other security risks if the health system’s staff were using a generative AI tool hosted in a public environment, such as ChatGPT, Dr. McSwain added.

“It’s within our system, so we control the information that goes into it. We can ensure that any information that is entered into it is kept confidential and secure,” he explained.

The new app is not UNC Health’s only foray into generative AI. The health system is one of the few that are piloting Epic’s new generative AI feature that allows clinicians to automatically draft message responses in the EHR. 

Epic announced this feature in April, when it said it was expanding its collaboration with Microsoft and integrating OpenAI services into its EHR. Other health systems piloting the tool include UC San Diego Health, UW Health and Stanford Health Care.

Photo: venimo, Getty Images