After nine months of research, a Washington State University sociologist concluded rural clinics aren’t ready for artificial intelligence implementation.
One of 25 fellows selected for the National Institutes of Health (NIH) AIM-AHEAD leadership program, WSU's Anna Zamora-Kapoor, partnered with Three Rivers Hospital in Brewster, Wash., to see if using AI-generated text messages and text-based intervention would help schedule screenings and follow up for patients who have an increased risk for lung cancer.
In her final presentation last month at the Washington State Academy of Sciences Annual Symposium, Zamora-Kapoor said rural clinics might not be ready for such AI implementation, after experiencing multiple barriers throughout the nine-month long project.
“We observed a lot of barriers in implementing this very simple study – barriers related to the clinic not being able to reach out to their patients,” Zamora-Kapoor said.
Zamora-Kapoor and her team at the Brewster clinic said they identified 144 eligible patients for the lung cancer screening. Using ChatGPT, they generated a “direct” and a “polite” text message to see which message would persuade patients to schedule an appointment.
After sending one of the text messages to each of the 144 eligible patients, eight patients who received the more direct text message scheduled an appointment. Only two of the patients who received the more “polite” message scheduled one.
The most shocking result, though, was the percentage of messages that bounced back.
“More than 50% of the messages that we sent bounced back from the patients, showing that the clinic doesn't have a way of contacting their patients,” Zamora-Kapoor said. “This is concerning in the case of if they need to coordinate an evacuation or something serious where they need to reach out to people quickly.”
Zamora-Kapoor said text messages bouncing back could happen for a number of reasons: lack of updated phone numbers, limited access to the internet and smartphones with restricted texting capabilities.
With the patients in the study being older than 50, it’s likely it’s a combination of all three.
Additionally, hospital Chief Operating Officer Jamie Boyer said she believes it could also be because of the large Hispanic community in the area.
“Because of where we are located, we have a really high population of Hispanic culture here,” Boyer said. “Some of the numbers are from Mexico or they're just a different kind of number so there are those challenges too, and we did get some of that.”
Although the study cited lack of communication as the main barrier, Boyer said the research project allowed the hospital to better understand how to use AI and expand its strategies to reach out to patients.
“We're not there yet, but we would love to be, and not every rural hospital and clinic are the same,” Boyer said. “Some have more technology than we do – so I think we're a little bit behind, but I don't feel like we're extremely behind.”
Similarly, Zamora-Kapoor said the text message project was an important starting point for the hospital. She hopes she can get another grant to focus more specifically on how the hospital could contact Hispanic and Latino residents.
“This study only touches the surface of the current challenges that clinics experience,” Zamora Kapoor said.
