Seattle health system testing AI tools to ease physician burdens, better patient care

Though AI’s wide implementation into healthcare is still in its beginning stages, a Seattle-based health system recently showcased how it’s using the technology to alleviate physician burnout and improve health outcomes at the HIMSS conference in Orlando.

“AI is not the magic bullet (but) it’s one of many many tools that we’ve got to (help address) this situation,” Aaron Martin, chief digital officer at Providence St Joseph Health, said on Monday.

The health system is currently testing an AI-supported virtual physician assistant tool and AI-based chatbot in an effort to help ease the burden on physicians and improve patient experiences. The virtual physician assistant helps physicians transcribe notes and has already reduced the amount of work physicians take home.

The AI-based chatbot is designed for patients and directs them to the appropriate venue of care based on their symptoms or conditions. It also offers FAQ-style questions. Though the chatbot has low click-rates, both the symptom checker and FAQ help pages have demonstrated high accuracy when routing a patient to right department or finding the right answer to a question.

With the technologies, the health system hopes to:

  • Collect more patient data to better inform physicians prior to visiting patients
  • Reduce or eliminate unnecessary care
  • Navigate patients to the right care options
  • Help direct lower licensed providers with conducting physical exams
  • Provide a consistent experience and continuity

“We think that this continuity across the entire continuum, it should know you regardless of what type of technology you’re using with AI and (machine learning), and it should have that context. It should follow you whether your not you’re using a smart speaker at home or you’re using the chatbot in or out,” Martin said.

Martin’s team within the hospital system focuses AI’s role within six digital journeys:

  • Making caregiving easier
  • How to better serve Medicaid
  • How to power behavioral health
  • How to make healthcare more personalized and convenient
  • Simplifying care
  • Creating new revenue streams

For AI tools to be effective they must be accurate, modular, context-aware, multi-channel and persistent in order to ensure encounters between physicians and patients remain focused on a patient’s needs, Martin said. When pursuing digital solutions, the health system’s model first requires it to define a specific problem and size it. It then attempts to see if it owns or can find a solution before setting out to develop and build its own digital tool to address a problem.

“Startups get very fixated on actual real problems in healthcare that matter but aren’t going to be the top priority for a health system, and they get frustrated because they don’t see their technology being scaled,” Martin said. “One of the things that we know as a health system in the technology community is to be very crisp about what our priorities are in terms of what are the problems we want to solve.”

Martin said the health system’s goals center around innovating at the end of the value chain—the creators and the consumers. In healthcare, that means centering solutions around physicians and patients.

“In any value chain, in any industry, there’s only two constituents that matter: there are the folks that actually create the value and the folks that actually consume the value,” Martin said. “In healthcare, there are folks that deliver and create value—clinicians and caregivers that see our patients—and there are folks that receive that care. If you are not working to reduce that friction between those two entities, you are in big trouble over the fullness of time.”