When it comes to weighing AI's use in healthcare, Dan Roth, MD, chief clinical and community division operations officer of Trinity Health, asks three questions.
In an industry full of static noise about new technology, Dr. Roth asks whether the AI candidate is clinically appropriate, resilient and promotes health equity.
"There's never been a more exciting time that I can recall in healthcare and a little bit of a disconcerting time around technology," Dr. Roth told Becker's on Dec. 1.
Two examples came to mind.
Livonia, Mich.-based Trinity is rolling out ambient listening, an AI platform that records conversations between caregivers, patients and their family members to generate immediate clinical notes. The application allows patients and clinicians more "face-to-face time," Dr. Roth said, as healthcare workers aren't flitting between computer screens and their patients.
The technology solves a current problem of patients feeling disconnected from medical professionals, and it saves caregivers time by reducing the heavy burden of documentation. On average, physicians in the U.S. receive three times more EHR messages than their counterparts across the globe, according to the American Medical Association.
Another example of Dr. Roth probing AI's clinical capabilities is predictive technology, which is growing rapidly. Predictive AI tools work to evaluate who, in a patient population, is most likely to become hospitalized, experience a safety concern or need specific treatments.
This technology's promise could either sparkle or fizzle.
"I have a pit in my stomach about it, and the pit can be both good and bad at the same time," Dr. Roth said. "It's amazing to think about the potential of this. It will transform how we do it. But I worry a lot. That same pit is me worrying about the potential negative repercussions of this."
Those repercussions could come in the form of worsened healthcare disparities, he said, since it's unclear how these technologies arrive at their conclusions.
"Many of these end up being sort of black box," Dr. Roth said. "It's hard to know how the tool arrived at its conclusion. That requires us to pause and discern on making sure the accuracy of it is going to endure as it's applied across different care settings and across the country or the world."