image: Predicting Missed Appointments in Primary Care: A Personalized Machine Learning Approach
Credit: American Academy of Family Physicians
Background and Goal: This study examined whether machine learning could predict the risk and contributing factors of no-shows and late cancellations in primary care practices.
Study Approach: Researchers at Pennsylvania State University integrated prior appointment history from 15 family medicine clinics, linking to corresponding U.S. Census statistics and national weather reporting databases. Four different machine learning modeling approaches, including gradient boost, random forest, neural network, and LASSO logistic regression were applied to predict appointment outcomes. The outcome of each appointment was attributed to one of the three classes: no-shows, late cancellations (canceled within 24 hours before appointments), and completed visits.
Main Results:
-
The analysis consisted of 109,328 patients and 1,118,236 appointments, including 77,322 (6.9%) no-shows and 75,545 (6.8% late cancellations).
-
The gradient boost model achieved the best performance in classifying patients as likely to be a no-show or to cancel an appointment late (AUROC of 85% for no-shows and 92% for late cancellations).
-
No bias against patient characteristics (sex and race/ethnicity) was detected.
-
The schedule lead time (the number of days from a patient’s appointment request to the appointment date) was the most important predictor of missed appointments.
-
Patients who missed appointments tended to be female, younger, sicker, under/uninsured, less fluent in English, and in ethnic minority groups. They also experienced longer lead times, higher prior missed appointment rates, and more socioeconomic challenges.
Why It Matters: The findings of this study provide insights into the underlying barriers to missed appointments and suggest that health systems prioritize strategies to reduce lead time and enable care teams to design personalized interventions, such as text reminders or transportation assistance to potentially improve patient appointment adherence.
Predicting Missed Appointments in Primary Care: A Personalized Machine Learning Approach
Wen-Jan Tuan, DHA, MS, MPH, et al
Department of Family and Community Medicine, Penn State College of Medicine, Hershey, Pennsylvania
Journal
The Annals of Family Medicine