AI is taking the healthcare industry by storm and has many important use cases for digital health companies, especially on the administrative side and with supporting clinical decisions. However, according to Omada Health, a virtual healthcare provider, there is one thing the company wouldn’t use AI for: spurring behavior change.
As a virtual care company for diabetes, weight management, hypertension and musculoskeletal issues, Omada is greatly focused on behavior change. But encouraging behavior change is better done by a human than by AI, said Dr. Carolyn Bradner Jasik, chief medical officer at Omada Health.
“Can someone feel accountable to AI?” Jasik said during an interview at ViVE in Los Angeles on Sunday. “So if AI is telling an app user, ‘Hey, this is a trend we see in your data’ or ‘This is what things look like,’ will a human patient take an automated suggestion and take action on that from a behavioral standpoint? I think it’s a really interesting question, especially now that AI has evolved so much. In the current state, we still believe that humans want to be accountable to humans.”
She added that when members communicate with Omada Health’s human coaches, they often ask if the coach is a bot because they’d rather work with a real person. Being guided by a real person provides the patient with more accountability and support, Jasik said.
That said, AI may evolve down the line.
“Maybe in the future, AI will get to the point where it can feel empathetic and supportive, but in the current state, our point of view is that humans still need humans to be accountable to and to get support from,” Jasik said.
What does Omada Health use AI for? Internally, it’s using the technology to streamline business operations and processes. That includes “some of our engineering workflows and clinically staying up to date on the latest evidence,” Jasik said.
The company is also looking at incorporating AI into its care for members and enhancing the work that providers do through clinical decision support. That includes providing data and suggestions to help providers be more efficient and offer more personalized care. This is something that would have been helpful when seeing patients, Jasik said.
“When I was in practice, one of the biggest challenges with a diabetes patient was trying to figure out how to match their treatment regimen with what they’re eating at home and what their usual habits are,” she stated. “It would take me a few minutes to look at their glucose data and their food log and kind of process it and think about what recommendations to make. AI is a great example where it can do some of that work, but then the human has to be the final one to sign off before they make the recommendation.”
Photo: metamorworks, Getty Images
