Annals of Family Medicine: Machine Learning Model Accurately Predicts No-Shows and Late Cancellations in Primary Care

PROVIDENCE, R.I., July 29, 2025 /PRNewswire/ -- Findings from a new study published in Annals of Family Medicine show that a machine learning model accurately predicts no-shows and late cancellations in primary care, providing insights for developing personalized strategies to improve appointment adherence.

The study, conducted by researchers at Pennsylvania State University, integrated clinical, geosocial, and environmental data from 15 family medicine clinics to personalize predictions of no-show, late cancellation, and completion of primary care visit appointments for more than 109,000 patients and more than 1 million appointments—including about 77,000 no-shows and 75,000 late cancellations. The team leveraged multi-class machine learning models, comparing four different approaches, including gradient boost, random forest, neural network, and LASSO logistic regression to predict appointment outcomes.

The gradient boost model achieved the best performance in classifying appointments as likely no-shows or late cancellations, with Area Under the Receiver Operating Characteristic curve scores (AUROC, a 0-to-1 measure of overall prediction quality; 1.0 is perfect, 0.5 is chance) of 0.85 for no-shows and 0.92 for late cancellations. A fairness check on the gradient boost model showed that predicted results were not biased against sex or race/ethnicity patient characteristics. Lead time—the number of days between booking and visit—was the most important predictor of missed appointments: lead times longer than 60 days were associated with a greater risk of missed appointments.

 "Given the strong effect of lead time, clinics could prioritize shorter wait times for high-risk patients," the authors write, and "machine learning-based analytics could help clinicians anticipate patient-specific needs, personalize outreach efforts, and proactively facilitate appointment scheduling."

A related special report in the same issue of Annals of Family Medicine proposes five high-level considerations around the data transformation that is needed to make studies like the one above more common. These considerations include the automation of data collection, organization of fragmented data, identification of primary care–specific use cases, integration of AI and machine learning into human workflows, and surveillance for unintended consequences.

For these efforts to be effective and work cohesively, it requires increased collaboration of industrial and academic AI and machine learning communities with primary care, increased funding from the private and public sectors, and upgrades to human and data infrastructures.

"These types of cross-sectoral collaborations are key to realizing the transformation of primary care data into a treasured resource that can unlock the true potential of artificial intelligence and machine learning in primary care," the authors write.

Articles Cited:

Predicting Missed Appointments in Primary Care: A Personalized Machine Learning Approach

Wen-Jan Tuan, DHA, MS, MPH; Yifang Yan, MS; Bidal Abou Al Ardat, MD, Todd Felix, MD; Qiushi Chen, PhD

Data Transformation to Advance AI/ML Research and Implementation in Primary Care

Timothy Tsai, DO, MMCI; Julie J. Lee, MD, MPH; Robert Phillips, MD, MSPH, Steven Lin, MD

Annals of Family Medicine is an open access, peer-reviewed, indexed research journal that provides a cross-disciplinary forum for new, evidence-based information affecting the primary care disciplines. Launched in May 2003, Annals of Family Medicine is sponsored by six family medical organizations, including the American Academy of Family Physicians, the American Board of Family Medicine, the Society of Teachers of Family Medicine, the Association of Departments of Family Medicine, the Association of Family Medicine Residency Directors, and the North American Primary Care Research Group. Annals of Family Medicine is published online six times each year, charges no fee for publication, and contains original research from the clinical, biomedical, social, and health services areas, as well as contributions on methodology and theory, selected reviews, essays, and editorials. Complete editorial content and interactive discussion groups for each published article can be accessed for free on the journal's website, www.AnnFamMed.org.

SOURCE Annals of Family Medicine