AI Tool Shows Promise in Improving Colonoscopy Adherence

Artificial neuron in concept of artificial intelligence: © ktsdesign - stock.adobe.com

A study presented at the 2024 American Society of Clinical Oncology (ASCO) Annual Meeting press conference suggests that using artificial intelligence (AI) virtual navigation could be an effective way to get people to complete colonoscopies. This approach was well-received by patients and led to higher appointment adherence rates, according to the quality improvement study.

“We found that MyEleanor, our AI virtual navigator, demonstrated high feasibility through high patient acceptance and engagement, clinically significant impact on reengagement, increased patient volume, [and the] ability to identify barriers,” said Alyson Moadel, PhD, deputy director of community engagement and cancer health equity, and the founding director of the Bronx Oncology Living Daily Program at Montefiore Einstein Comprehensive Cancer Center in the Bronx, New York, during the presentation.

Moadel and her team chose to focus on colorectal cancer screening due to the biggest disparities both in the Bronx and the United States altogether. She noted that African Americans have a 20% higher incidence of colorectal cancer and are 40% more likely to die from the disease. In addition, early-onset cancers often affect Latin populations, which makes up more than half (52%) of people living in the Bronx.

“We were really focused on how can we engage patients who are not showing up or canceling their appointments, and we had quite a lot,” Moadel said. “As a psychologist, I'm also interested in barriers to screening. We can get people in, but if we don't understand what their barriers are, we aren't going to be as successful as we want to [be].”

The team decided to leverage AI to call 2400 patients (mean age, 57.50 years; range, 31-85) who were nonadherent with their colorectal screening between April 2023 and December 2023. The population consisted of Hispanic (44%) and Black (39%) patients, 64% women, and both English (73%) and Spanish (24%) speakers. Some patients in this population were also unemployed (32%) and married (43%).

MyEleanor, the English/Spanish-speaking AI virtual navigator, called the patients to discuss rescheduling, assessed barriers to colonoscopy uptake, offered live transfers to clinical staff to reschedule their appointments, and provided procedure preparation reminder calls.

To make the AI navigator seem more patient centered, Moadel and her team developed the following script when the technology would call patients: “I’m Eleanor, the automated care assistant for your team at Montefiore. I am calling today because we noticed that you missed your most recent colonoscopy appointment, and we would like to help you reschedule it at the end of this call.”

With regards to engagement with MyEleanor, of the 2400 patients who were called, 57% (1368) engaged with the call.

“I actually listened to a couple of [these calls] last night, and [the patient] would speak [to the AI navigator] sometimes up to 9 minutes,” Moadel said. “The average was about 6.5 minutes.”

More than half of the patients (58%) who were called by MyEleanor accepted a live transfer to reschedule their appointment, and 25% of these patients completed their colonoscopy. The no-show completion rate also nearly doubled from 10% to 19%, and patient volume for colonoscopies increased by 36%.

“So that was another 400 or 500 patients that our patient navigators—who saved something like 52 hours a month—were able to get the new patients through, so that was super exciting,” Moadel added.

To learn more about the barriers to colorectal cancer screening, MyEleanor would ask patients during the call whether they would be interested in answering a few questions about what stopped them from coming to their colonoscopy appointment. These questions focused on 14 different barriers, from social determinants of health (including social needs and transportation) and COVID-19 concerns, among others.

Thirty-eight percent of patients reported that transportation was the barrier that stopped them from attending their colonoscopy appointment. “That’s often because they needed an escort or they just couldn’t afford it,” Moadel said.

Other barriers that patients reported included perceived need for a colonoscopy, medical mistrust, and fear of the findings. For the patients that accepted the transfer to reschedule their appointment, they often had a higher number of barriers; for example, patients who were unemployed or disabled were more likely to accept the transfer. Patients less likely to accept the transfer to reschedule their appointment reported greater concerns around cost, medical mistrust, or cultural concerns like those around language, immigration, or religion. Those who were less likely to complete the colonoscopy procedure also had concerns around cost and fears about the procedure and its findings.

“That gives us an idea of how we can address these either through MyEleanor or through other peer navigators and cancer survivor navigators that we also have at Montefiore,” Moadel said.

She added that Spanish-dominant patients and those who declined to identify their race had the greatest number of barriers compared with the rest of the population (P for both < .001)

“That was also very informative,” Moadel said. “This we have to dig in more deeply to kind of understand, but this gives us a sense of some of the psychosocial nuances related to patient engagement.”

The next steps for Moadel and her team include measuring the impact of using AI on patient navigator burden, patient satisfaction, other screening programs such as lung and breast, distress screening, and cost savings.

Fumiko Chino, MD, radiation oncologist at Memorial Sloan Kettering Cancer Center in Middletown, New Jersey, commented on the findings of the study during the press conference.

“The conversational AI navigator was successful at engaging patients and essentially doubled the colonoscopy completion rate for this at-risk group of patients who had previously no-showed for a colonoscopy,” she said. “A key aspect to highlight is that the patients who conversed with MyEleanor were on average in their 50s, and three fourths were Black, Latinx, or Hispanic, so the intervention really did work in the population at highest risk for screening gaps.”

She added that the benefits of the AI navigator will need to be continually reevaluated over time and tested in other populations, especially since this study was a quality improvement study and not a randomized-controlled trial. Despite this, she said that this initiative may have several benefits for both patients and health care systems.

“Overall, this quality improvement initiative is a truly innovative means of increasing cancer screening,” she said. “It really offloads the work from an overburdened health care workforce by leveraging artificial technology to optimize the outreach capacity to vulnerable populations. The work really delivers on the promise of technology to facilitate better, more efficient, more equitable cancer care, and I really anticipate that it will ultimately improve cancer outcomes.”

Reference

Moadel A, Galeano D, Bakalar J, et al. AI virtual patient navigation to promote re-engagement of U.S. inner city patients nonadherent with colonoscopy appointments: A quality improvement initiative. J Clin Oncol. 2024;42(suppl 16):100.