Artificial intelligence has promised to make doctors’ lives easier. It’s not the first form of technology to do so—which has invoked some skeptics.
The electronic health record (EHR), which digitized patients’ medical histories, also aimed to reduce physicians’ burdens. It eliminated the need for handwritten notes, but it deepened the well of available data on each patient. That data can be hard to navigate as a doctor, according to Dr. Pete Clardy, senior clinical specialist at Google Health.
Dr. Clardy participated in a September 17 panel at Newsweek’s Global Headquarters in New York City. He was joined onstage by Dr. Ashley Beecy, medical director of AI operations at NewYork-Presbyterian; Dr. Eric Williamson, associate chair for radiology informatics at Mayo Clinic and supervisor of its Radiology Artificial Intelligence program; and Dr. Christine Sinsky, vice president of professional satisfaction at the American Medical Association. All four panelists are medical doctors themselves.
The amount of available health and wellness data has increased substantially over the past 20 years, Clardy said. Since health data is so complex, it suffers from a “fragmentation problem” and gets scattered in different places and different formats.
Some new AI tools aim to sort and summarize that data for doctors. But health care is sometimes perceived as slow to adopt new technology, according to Clardy, a pulmonologist and critical care physician. He cited two reasons why doctors might resist the “shiny new thing.”
First of all, “there’s a lot of risk when [new technology] is involved in the care of other people,” Clardy said.
Plus, physicians’ time and attention live on “razor-thin margins,” Clardy continued. “Anything that breaks a workflow is going to be resisted, even if it might lead to a better workflow.”
It doesn’t help that EHRs think differently from doctors. Medical records are sorted by data type, but clinicians prefer to analyze information that’s organized around specific concepts or diseases.
There is plenty of opportunity for AI to develop side by side with the EHR, Clardy said. He recommends that AI tools nail down the specific issues they’re trying to solve for. Otherwise, they risk complicating processes that are already complex.
If AI doesn’t cite specific goals, it could cross the line from an “information organization tool” to a “medical device,” according to Clardy. “It’s very easy with new technology to make a medical device without realizing it.”
Health care can learn great lessons from the EHR, that first major wave of innovation in health care technology, according to Sinsky of the American Medical Association. During the “hype phase” of the EHR, health care stopped putting effort into other systemic problems because the industry believed the new technology could fix everything.
Sinsky foresees the same thing happening with AI, but on a greater scale. People might neglect preexisting issues, like doctors’ overloaded inboxes, for hope that AI can solve the problem.
“Technology isn’t always the solution to the problems that have come from technology or are associated with it,” Sinsky said.
Read more about AI and its potential implications for doctor burnout here.