Artificial intelligence had a rapid rise in health care, but one health system AI chief says treating the technology as a new phenomenon could cause the industry to overlook decades of lessons about how to use it. Philip Payne, chief health AI officer at BJC HealthCare and WashU Medicine, pushed back on the characterization of AI as an emerging field during Newsweek's AI Health Summit in New York City. AI research does stretch back decades, while its presence in everyday medical practice has expanded as computing power, data availability and technical infrastructure have advanced.
Payne himself was discussing AI applications in clinical practice years before the generative AI boom. In 2018, he pointed to computerized identification of abnormal heart rhythms as a simple form of AI already operating in clinics. It wasn’t until late 2022 that most health systems started to seriously consider deploying increasingly capable forms of the technology across entire organizations.
Rohit Chandra, executive vice president and chief digital officer at Cleveland Clinic, described today's technology as “probably the most immature version that we're going to see.” That rapid evolution, he said, means organizations need room to experiment and accept that some efforts will fail. That journey, the panelists argued, requires health systems to continuously learn on two fronts: from what decades of research have already established about AI in medicine and from what increasingly capable technology makes possible now. One of the questions organizations are still working through is what AI needs around it to deliver on its potential.
Deploying more capable models across an organization requires infrastructure capable of feeding them useful information. Bilal Muhsin, executive vice president and president of the connected care segment at the global medical technology company BD, pointed to data as one of those constraints. Health care, he said, needs continuous, real-time information to support increasingly predictive applications.
Without continuous data capable of feeding AI models, he added, “we will all fail.” As those capabilities advance, health systems are also learning what becomes possible when AI has access to information closer to the moment care is delivered. At Houston Methodist Hospital, some of that infrastructure is already supporting clinical care. Michelle Stansbury, associate chief innovation officer, said continuous vital-sign monitoring has allowed clinicians to identify potential problems earlier.
In one case, she said, staff in the health system's virtual operations center detected a concerning signal and alerted a bedside nurse. For Payne, expanding technical capability also raises the stakes for how health systems measure whether AI works. He warned against focusing heavily on process measures simply because they are readily available, telling the audience of health care leaders that, “Outcomes are everything.” At BJC and WashU Medicine, Payne said his program will not undertake an AI project without a clear pathway toward measuring outcomes.
Extract — continue reading at the source.
