Why artificial intelligence underperforms hospital patient engagement
New research shows that health systems are struggling to effectively use artificial intelligence to improve patient engagement.
The study, released earlier this month, found that while investment in tools such as environmental scribes is booming, artificial intelligence applications for patient engagement are lagging. For the study, patient engagement startup Lirio commissioned healthcare consulting firm Sage Growth Partners to interview more than 75 healthcare system executives across the U.S.
Only 5% of these executives said they were satisfied with the tools they have to address common patient engagement challenges such as medication adherence and missed appointments, which not only lead to poor health outcomes but also cost billions of dollars in avoidable medical costs each year.
To help hospitals address these gaps, companies selling patient engagement tools must move to an “N-of-1 personalization” model, said Lirio chief behavioral officer Amy Bucher.
“In health care, the standard approach to personalization is not very personal,” she declares.
Typically, personalization starts and ends with certain form fields, such as name or age range. Bucher points out that these methods essentially just classify people based on demographic data, rather than taking into account their individual motivations and behaviors.
For example, a provider might send the same generic email reminder about mammogram services to all women age 40 and older. But not every woman in this broad age group needs the same type of information, Bucher explained, and the N-of-1 approach goes a step further to generate customized information that takes into account each patient's unique needs, behaviors and barriers.
“If a woman hasn't had a mammogram in a few years, N-of-1 personalization considers why that's the case. Is it hard to fit an appointment in with work? Does she need child care? Hate the anxiety that comes with cancer screenings? Whatever the case, personalization that doesn't address this won't be as effective,” commented Bucher.
She added that recent advances in artificial intelligence create the ability to scale N-of-1 personalization.
She points out that humans are able to do this very well on a 1:1 basis – we can take in complex information from what people are telling us, as well as non-verbal signals and contextual clues, and adapt our approach on the fly. But Bucher said humans can't scale across large patient populations and can't afford to use real-time support for every use case.
“Technology has long been able to process more complex and larger data sets than humans, but only recently, with the explosion of agent AI and the use of techniques like reinforcement learning, has it also been able to produce meaningful N-of-1 output,” she said.
She also noted that better personalization can bring new levels of efficiency and connection.
Take diabetes for example. The disease affects one in 10 Americans, but despite its prevalence, people with diabetes often don't connect with health care providers, Bucher said.
“The standard approach of trying to get people to schedule appointments and take their medications is clearly not going to work for everyone. Personalizing these outreach campaigns can help spark interest and get people thinking differently about the value proposition of taking action, while operational efficiencies can be improved through digital outreach,” she commented.
The diabetes use case highlights why patient engagement may be one of the most promising and underutilized AI applications in healthcare. Bucher claims that when personalization goes beyond demographics and addresses individual barriers to mobility, it can not only drive clinical improvements but also help health systems engage patients at scale.
Photo: Paul Bradbury, Getty Images
