HCA Healthcare pilots Google AI to aid handoff documentation

Nurses at two southern HCA Healthcare hospitals are testing out Google’s generative artificial intelligence (AI) technology to help make handoffs more seamless and cut down on administrative burdens, executives announced Tuesday.

The tech, which is deployed on a hospital-issued iPhone, tracks a 12-hour shift to create a summary and task list for dozens of nurses at TriStar Hendersonville Medical Center just outside of Nashville and UCF Lake Nona Hospital in Orlando—two hospitals in HCA that are dedicated “laboratories” for various innovation efforts, Michael Schlosser, HCA’s SVP for care transformation and innovation, told Healthcare Brew at Google Health’s annual event, The Check Up.

A nurse preparing to end their shift can then review the MedLM-created breakdown and make edits to it before transferring that “critical information” to the next incoming nurse in a handoff—a process that’s typically lengthy and manual, Schlosser said during a presentation at the company’s Pier 57 office.

“Apart from noting the standard data like vitals and allergies, [the handoff report] also relies on the nurse’s recollection of events and conversations, and data points that they consider relevant to the transition, like how the patient slept or how often they asked for medications,” he added.

Kyle Brashear, director of clinical design and care transformation and innovation at HCA Healthcare and a nursing operations manager at Orlando Health, added in a video shown at Google’s presentation that when patient information isn’t effectively shared, nurses risk creating poor patient health outcomes.

Though Google’s generative AI for nurse handoff reports is currently only used in med-surgical (med-surg) units, which are beds reserved for general hospitalizations and HCA’s most common unit type, Schlosser told Healthcare Brew he expects to roll out the technology to more hospitals in the coming months.

“It’s also probably the area of the hospital that gets the least love just because it’s a generic med-surg unit. [There] tends to be a lot of really new nurses, though, so they could really use the help,” he said, adding that the system hired about 12,000 new nurses in 2023. “It’ll take a little development work to go to other units, like to ICUs or labor and delivery, because the way they do handoff is a little different.”

The health system, one of the largest hospital operators in the country that does 400,000 handoffs a week, has explored other Google technologies to automate documentation and summarize medical records insights. Schlosser said the long-term goal of using the large language model in the hospital is to help the AI “think like a nurse.”

“That’s a reproducible capability […] we can then take and use all over the place for other types of handoffs, for discharges,” Schlosser told Healthcare Brew. “So the handoff by itself is a great use case, but we picked it and we’re excited about it because it’s the starting point for something that’ll take us much further.”