AI in LA: ViVE Gives the Starring Role to Workflow Tech

Everyone's talking about gen AI, but how is it actually being used?

While Hollywood may be all abuzz at the quickly rising fortunes of Barry Keoghan, generative Ai is seeing a similar ascendancy in healthcare. But will both avoid the bombs and tabloid headlines and fulfill their potential?

As ViVE 2024 kicks into gear today at the Los Angeles Convention Center, AI is on everyone’s minds. And while the past year has seen stories ad infinitum on the promise for large language models and gen AI to rid healthcare of its biggest headaches, the talk on the floor is that it’s time to show everyone what it really does.

“Every health system is investing a great deal in AI solutions,” says Harjinder Sandhu, chief technology officer for Microsoft’s Health and Life Sciences Platforms and Solutions, a partner for many health systems putting AI to the test. “You have to. There’s really no choice. But readiness really depends on the complexity of the use case and the risk of the use case.”

“What I see is a lot of caution,” he adds. “[Health systems] are being very tentative in thinking about these use cases. I see a lot more confusion than I see jumping too fast right now.”

[Also read: How to Create a Game Plan for Assessing AI Readiness.]

Sandhu, who will be part of a ViVE panel this week discussing AI readiness, says healthcare leadership is understandably cautious in moving forward with projects that carry a lot of risk, especially as the industry is still trying to figure out governance. But that doesn’t mean they haven’t come up with plenty of ideas.

“I’ve literally been in sessions where, over a four-hour period, you’ll have a group that puts Post-It notes on a board with, like, 300 use cases,” he says. “And [they’ll] say ‘Here’s all the different areas we want to be able to impact.’”

Some of the biggest, brightest Post-It notes are focused on using gen AI to capture conversations and turn them into valuable clinical information in the EHR. At a time where every healthcare organization is dealing with a shortage of clinicians and a surge in stress and burnout, it’s those tasks that are causing the most conflict. Those in the know call it “pajama time,” as in the time spent by clinicians each evening at home going over their notes from the day’s patient encounters and translating them into information they can use for care management.

Microsoft and Nuance recently rolled out DAX Copilot, which integrates directly into the Epic EHR. It’s one of several tools from a number of AI vendors aimed at that particular pai n point.

“It’s probably one of the fastest growing products that we have witnessed in terms of how quickly physicians are taking to it and adopting it,” Sandhu says of the market in general. “It is starting to make an enormous difference in how physicians view their work and their work-life balance.”

This particular tool captures the conversation right after it has taken place, giving clinicians their notes immediately after the patient encounter, while the conversations are still fresh. They can review those notes for accuracy, then submit them into the EHR.

[Also read: Study: Ambient AI Scribes Are Good, But Not Yet Ready For Pime Time.]

That review process is crucial. The technology is still relatively new, and still liable to make mistakes. That’s why practically any use of AI in healthcare at this time needs a “human in the loop” to review and sign off on the final product.

“You have to approach [AI tools] always with a hint of skepticism,” Sandhu points out. “Be a little bit skeptical about what they produce and double-check and triple-check.”

But the benefits are significant. Any tool that can integrate easily into a clinician’s workflow and reduce translational tasks—especially during nights and weekends—pulls time away from the computer and puts it back where it should be: In front of the patient or the family. Sandhu says those tools should see the lion’s share of adoption over the next year or two, especially as clinicians test them out and vendors work to fine-tune the process. It’s worth noting that gen AI is designed to learn as it goes along, so that a tool will learn a clinician’s habits and language and become better at transcribing.

Beyond that, Sandhu says healthcare decision-makers are keen to apply gen AI to another crucial pain point: Nursing workflows. Nurses are struggling just as much, if not more than, any other healthcare provider, and they need AI to reduce that overload and put them back in front of patients. But they also need technology that is designed for them.

[Also read: How to Create AI Programs With Clinical ROI.]

Even farther out, Sandhu sees an expanding market for gen AI technology that can capture and, more importantly, analyze conversations. Consider a tool that that study the patient encounter for signs of mental health distress, enabling specialists or even primary care providers to identify patients in need of help just by how they talk to someone.

“We’re kind of in this new age of literacy with AI,” he says.

So at this point, as the attendees at ViVE settle in under a crisp California sun, a lot of talk will be about what’s on stage now, and how it’s playing in the market, and not so much about that next big thing or the future blockbuster. It’s nice to see where this will all go, but there’s a view among the executives that new ideas have to show ROI now, not later. Healthcare needs help now.