For Rural Health Systems, AI is All About Saving Clinicians

Idaho's St. Luke's Health is embracing ambient AI not for the clinical benefits (just yet), but because it helps them keep their doctors and nurses.

Small health systems like St. Luke’s in Idaho aren’t embracing AI for the clinical outcomes (though that is a big benefit). They want to help their doctors and nurses, because there aren’t a lot of them to go around.

Provider stress and burnout is a concerning issue for rural health systems, with much of the grief tied to documentation and administrative duties. That’s why healthcare leaders are embracing ambient AI in droves even before the technology has proven its ROI. They want relief now, and will map out the benefits later.

“Our CFO said, ‘Look, we want to make a positive impact on provider well-being,” says Reid Stephan, VP and chief information officer of the Boise-based, six-hospital, not-for-profit network. “If that's all we do with this pilot, I will consider it a success.”

By all accounts, that plan is working.

Working through their Epic EHR, St. Luke’s installed Ambience Healthcare’s AI scribing tool last April, and Stephan says they’re now seeing results. Over the past year, the health system has seen a 38.8% decrease in overall documentation time for clinicians, including a more than 40% drop in after-hours documentation. This translates to a 22% increase in face-to-face time between clinicians and their patients and a 25% reduction in clinician burnout.

Stephan says those numbers translate to happier providers, a key factor in a region where the doctor-patient relationship is as durable as the Rocky Mountains surrounding Boise, a city of some 235,000 people. The idea behind using AI isn’t necessary to increase access to care, but to enrich those care pathways.

“Our approach has been we're not going to pursue supply-side-driven AI opportunities because the amount of supply-side opportunities is enormous and we might get lucky, but more likely we're going to just waste resources and capacity hoping something's going to work,” Stephan pointes out. “So we really focused on the demand-side-driven need we have. And again, it was an obvious one: primary care, in particular. Can we use generative AI specifically to help then with that patient-provider encounter?”

Because of its size, Stephan says the health system isn’t looking to be a trailblazer with AI, but rather wants to use the technology to address its specific pain points.

“We aren’t pioneering this,” he says.

Reid Stephan, VP and Chief Information at at St. Luke's Health. Photo courtesy St. Luke's Health.

There are financial benefits as well. St Luke’s is using a “coding aware” platform, which not only captures the conversation but provides real-time coding of the encounter. The tool is reportedly generating more than $13,000 annually per clinician through more accurate coding and better communications, and Stephan says the deployment paid for itself within five months.

Whether those numbers play out over time remains to be seen. Critics say these AI tools are great at catching early benefits, but long-term results are hazy. Stephan, on the other hand, says as long as his clinicians are happier and engagement is better, the value is there for him.

He also says the results of the first year of implementation have given leadership the support to expand the platform. After starting with family medicine clinicians, the health system is now using the AI tool in some 28 specialties.

“This instilled in us a lot of confidence that we can do this again and again and not have to recreate … across each different specialty,” he says.

In fact, Stephan says he was surprised at how fast clinicians caught on to using AI, and would have considered expanding the pilot sooner had he known the results.

“The word spread across the community, and within a few days or a couple of weeks of the initial rollout of the pilot, we started to have a groundswell of hands being raised,” he says. “’When is my turn? Why wasn’t I part of the pilot? How can I get this sooner?’”

Stephan says it’s important to temper expectations with AI, not only to make sure the hype doesn’t overtake reality but to make sure clinicians know what they’re using. While he tells them AI will improve their workflows, that doesn’t mean management wants them to take on extra patients or patient visits or do more work.

True ROI, he says, comes in a more enriched doctor-patient encounter, where both doctor and patient are more engaged.

“Maybe then the provider is able to pick up on something or hear something  that they might have otherwise missed when they’re trying to do the swivel chair game of typing and listening,” he says.