A team of MetroHealth and Case Western Reserve University researchers used artificial intelligence to improve clinic access for Black patients.
Dr. Yasir Tarabichi, medical director of the virtual care enterprise and director of clinical research informatics at MetroHealth, and a team published a study earlier this year that showed how using AI to predict the probability of no-show appointments combined with personal outreach to at-risk patients could improve show rates, particularly among Black patients.
“The framing here for this particular work is, there’s a lot of algorithms, predictive algorithms and artificial intelligence, machine-learning algorithms out there, that can show you the risk of an event,” Tarabichi said. “In this case, we’re able to predict the likelihood that somebody’s going to miss their appointment.”
No-shows negatively affect patients and caregivers alike. For a patient, missing an appointment means losing access to crucial health care. For a physician, it results in wasted time that could be spent taking care of someone in need. Tarabichi and his team looked to use AI to address this issue, with the understanding that AI can have negative consequences if not used properly.
To prevent biases, Tarabichi and his team started by asking themselves questions. Is there value in doing this work? Will the predictions be fair and equitable to diverse patient populations? What can be done with the information gathered? Will the actions that follow also be fair and equitable?
With no-shows, it was important to consider the factors that led to patients missing appointments. Tarabichi said transportation and childcare barriers tend to prevent at-risk patients from making it to their health care provider. But simply using AI to predict the probability of an at-risk patient missing an appointment and then giving that time slot to someone else would only worsen health inequities.
“If we just applied something without thinking, if we just decide somebody’s not going to show up, we’re going to open up that appointment for somebody else, for instance, that’s a reallocation of scarce resources, which could actually widen disparities,” Tarabichi said. “If you give people who are more available more opportunities for scheduling, then you start edging folks out.”
Instead, the researchers used the predictions to provide additional outreach to potential no-show patients. A random selection of these patients received phone calls from MetroHealth schedulers between January and September 2022. If patients said that transportation or other barriers might prevent them from making it to their appointments, then schedulers offered resources or telehealth options. Black patients who received phone calls saw a 36% reduction in no-show rates compared to those who did not receive a phone call.