AUSTIN, Texas — Automating phone calls with patients may make economic sense, researchers said here.
A clinic’s new system of automated appointment reminder phone calls was associated with a decrease in late cancellations of less than 48 hours notice (23.9% to 17.1% after implementation, P=0.002), reported Joel Raborn, MD, a radiology resident at the University of Alabama at Birmingham, at the Society of Interventional Radiology (SIR) annual meeting.
There were no significant changes in arrivals nor early cancellations, however, and no-shows actually trended upwards though not significantly (6.3% vs 13.7%, P=0.06).
The study was based on a retrospective review of appointments before and after the interventional radiology clinic started using the automated system. The phone calls were sent in a series, on top of another reminder letter mailed 2 weeks before the appointment.
Established in 2016, this clinic is fairly new and saw visits go up from 177 to 209 a month before and after the automated phone system was in place (P=0.0001). Moreover, revenue increased from $22,730 to $30,000 per month (P=0.0001), according to Raborn.
Another SIR presentation took it one step further, showing how machine learning technology has the potential to interact with patients to accomplish the goals of the pre-procedural telephone call.
Kevin Seals, MD, a fellow in interventional radiology at the University of California, San Francisco, demonstrated with a video how his group’s human-sounding chatbot could ask a patient about the presence of allergies and sleep apnea, the use of blood thinners, whether the patient has transportation to go home after the procedure, and any questions he or she might have.
“This is just a baby step in a range of fascinating things we can do in the future,” Seals told the audience, noting that machine learning in radiology can have applications outside just image processing.
The researchers set up the robocaller using Google’s open-source machine learning software Dialogflow, inspired by a May 2018 demonstration of this technology making a real phone call to book a dinner reservation.
Benefits of these phone calls in healthcare include cost-effectiveness and the chatbots being “super consistent” (not making human errors) and able to be programmed to follow evidence-based guidelines for patient contact.
Moreover, robocalls are a good way to reach elderly patients, who may not always want to use an online web portal to get lab results and other information, Seals said.
“Machine learning has the prospect of improving many of our processes, reduces the need for people to do them, and increasing their accuracy and efficiency,” commented Harlan Krumholz, MD, of Yale School of Medicine in New Haven, Connecticut, who was not involved in the studies.
“The key will be to develop systems that people find useful — and even delightful, which will require a focus on user-centered design. The best systems will seem helpful, personalized, and timely — and people will not even know that there is technology involved,” he said.
Seals disclosed no relevant relationships with industry.