Physicians at Southwest Michigan Health System See Success With Ambient Speech Technology

The health system’s CMIO calls the conversational AI technology a 'game changer in healthcare.’

A health system in southwest Michigan is betting big on conversational AI technology to improve care and the doctor/patient experience in primary care.

University of Michigan Health-West has implemented Nuance’s Dragon Ambient eXperience (DAX) technology for its primary care physicians, and its executives believe conversational AI technology such as this to become the standard of care within a few years.

The benefits, executives say, include expediting the prior approval process for referrals, which contributes directly to the health system’s bottom line.

"We're doing a systemwide primary care roll-out of this," says Lance Owens, MD, chief medical information officer at University of Michigan Health-West.

During a pilot test of the technology with 15 primary care physicians, the health system's auditors found that DAX directly led to a 40% increase in first-time approval of prior authorizations "because the documentation is so good," Owens says. "It's a life preserver for healthcare."

Capturing the Conversation

DAX works by capturing conversations between physicians and their patients and outputting a SOAP note generated by DAX's algorithms. These notes find their way into each patient's electronic health record in an hour or less, according to Owens.

The technology addresses a major complaint that physicians have had with EHRs ever since they were implemented: Many of them feel the technology makes them more like highly-paid scribes and recordkeepers than doctors.

"It essentially eliminates that cognitive burden of having to put that information into the record," Owens says. "I can now, for just about any visit, walk into an exam room, sit down, turn on DAX, and then literally sit back, look my patients in the eye, have a meaningful conversation, be attentive, being able to synthesize that information, formulate a plan, talk about my medical decision making with the patient, and do my patient education. And knowing that all of that is somehow going to make it to the record in about an hour or less is an absolutely amazing thing."

Setting up the technology for additional physicians requires five minutes per physician to download the DAX app to their iPhones.

DAX's algorithms are able to distinguish different uses for words with more than one meaning, Owens says, such as distinguishing between a fall – as in a patient stumbling – and the fall season.

DAX also prompts physicians to enter orders and prescriptions as part of generated notes.

"If I'm talking to a patient about their X-ray, DAX just intuitively knows [to] show the X-ray" in the EHR, Owens says. "If I'm talking to a patient about starting them on hydrochlorothiazide for blood pressure, the order is queued up for me for signature. If the patient says ' My father had a heart attack since the last time we met,' DAX knows to put coronary artery disease in the family history of her father."

At present, Nuance sends each generated note to a quality data specialist, who reviews the note for accuracy and completeness. But the roadmap for the technology includes a completely automated process, according to Kenneth Harper, Nuance vice president and general manager of healthcare virtual assistants and ambient clinical intelligence. The company began testing those capabilities in December 2021.

"We can start bringing in clinical decision support into the workflow," Harper says. "We'll be able to allow the care teams to engage the patients even more productively, and make sure that the right patient outcomes are being delivered. You could just never do that with a human-based scribe type of system."

Harper says DAX currently saves providers seven minutes of documentation time per encounter.

"That typically equates to two to three hours of documentation time a week," he says.

Health systems using conversational AI technology could simply pack more patients per day into a physician's schedule, but many are using it instead to improve patient experience and working conditions for physicians, and to combat physician burnout.

"Our new vision statement is, ‘Our innovations change care and our care changes lives,’" says Josh Wilda, executive vice president and chief information officer at University of Michigan Health-West.

"Especially with the mission of healthcare being such a volatile mission these days, we got away from that message [of seeing more patients] and really made the message more about the ROI of provider well-being, of provider-patient intimacy, as well as the ongoing innovation of what this is going to bring to our team," Wilda says.

The health system turned to its charitable foundation to pay 60% of the first-year costs of the technology, Wilda says.

In terms of return on investment, Wilda says the technology gives them the ability to retain providers for many more years than might be possible without it.

"It's hard to put … a price tag on that kind of relationship," he says.

Conversational AI is Still a New Concept

Conversational AI is still in its infancy, due to the various challenges that face the clinical documentation process.

An effort to expand the use of DAX to specialists faces more challenges, in part because of the tendency of some specialists to perform "pre-charting," or carrying over written notes from previous patient visits into new patient encounters.

“I just don't think it worked well with our providers,” Owens says. “We're hopeful that as we grow this out, our specialists are just going to go, ‘Wow, when do I get that?’ There is no question in my mind this technology is going to be universal throughout the world."

"I wish from an industry standpoint, there was a golden standard for documentation, because actually, the AI would be able to learn that much faster," Harper adds.