How Wisconsin doctors and patients are using AI to assist with healthcare - WPR

More than 80 percent of physicians nationwide report using AI in their practice
More and more Wisconsin healthcare providers are turning to artificial intelligence.
A recent survey from the American Medical Association found that more than 80 percent of physicians now use some form of AI in their practice, more than double the rate from a 2023 version of the same survey.
The most common uses they reported were for taking notes and summarizing medical research, but some use it to assist with diagnosing from test results or drafting messages to patients.
WPR’s “Wisconsin Today” newsletter keeps you connected to the state you love without feeling overwhelmed. No paywall. No agenda. No corporate filter.
Dr. Brooke Crotty is an internal medicine physician and interim president for Inception Health, the innovation arm of Froedtert ThedaCare in Wisconsin.
She told WPR’s “Wisconsin Today” that AI note-taking, called ambient scribe technology, helps doctors and nurses be more attentive to their patients and spend less time focused on typing into their computers.
“The new AI scribes are able to be unobtrusive and invisible in the exam room, to take notes and help provide the clinician with a summary of the medical encounter,” Dr. Crotty said. “We’ve seen studies where we’ve seen more eye contact, and generally it has really helped clinicians and patients feel that it’s helping them.”
Health systems are also using AI agents to connect with patients between visits.
Dr. Crotty said the technology can do outreach to schedule patients for tests or screenings they might need, or it can touch base with a patient after a procedure or hospitalization.
“As a patient, you may be called by an AI agent who’s there to check in on you, ask questions, make sure that you’re recovering or doing well according to your care plan,” Dr. Crotty said. “If anything seems out of ordinary or off track, they can then route that to a human care team member who can then pick up the connection.”
The AI technology is not perfect. The two biggest concerns that Dr. Crotty watches for are AI hallucinations — when the technology provides wrong information — and AI sycophancy, which is when a chat bot is too agreeable and takes the user’s side when it should be disagreeing and pushing back.
She sees the technology continuing to improve in these areas, but it’s why she believes these tools need to be used with the guidance of a professional caregiver and aren’t here to replace clinicians.
“I think we’re going to co-evolve with AI over the next decade or so, where AI is part of the care team and part of the care process, but it is just that: It’s part of a broader process. It’s not a standalone service,” Dr. Crotty said. “I think the goal of any health care information technology is that it improves the health of the people in our population, and it keeps them safe.”
When people seek treatment advice or other health information from artificial intelligence without guidance from a clinician, that raises its own set of concerns.
University of Wisconsin-Milwaukee healthcare informatics professor Lu He studies how patients interact with AI.
She told “Wisconsin Today” that a patient might input their clinical lab results or other health data to a large-language model like ChatGPT and ask it for a second opinion.
“There are so many concerns,” Lu He said. “For example, the first is definitely, ‘What if the AI makes mistakes? Who will be responsible for those errors that may lead to severe consequences?’”
She also pointed to data privacy concerns, with patients voluntarily providing sensitive health information to this technology that isn’t required to protect data the same way healthcare providers do.
Lu He noted that infrastructure is an important part of the conversation too, with not every clinic having the human or technological capacity to properly implement AI tools.
“Hospitals in rural areas, in underserved areas, they’re way behind in terms of AI adoption,” she said. “Another thing is the patients, their own digital health literacy and whether it is enough to support their safe use of AI tools.”
Researchers like Lu He haven’t been able to match the rapid pace that AI is developing. Proper studies about the use of this technology in healthcare take a long time, and she said these tools have gone into place before independent scientists have fully fleshed out their efficacy.
Lu He hopes to see healthcare providers put more emphasis on teaching patients best practices for using AI, while the research tries to catch up.
“The first thing I observe in my own research is that patient education is very understaffed,” she said. “I think if there are enough patient educators, they should provide adequate knowledge and training to patients who are navigating this increasingly complex health system.”