Predicting Infant Nonattendance at the Next Recommended Well-Child Visit: Model Development and Validation

BackgroundWell-child visits (WCVs) are essential for preventive care, yet missed appointments often lead to delayed interventions. We developed and validated models to predict next-visit nonattendance using routine electronic health record data. MethodsUsing data from two Chicago-area pediatric practices, Practice A (1,215 patients; 3,654 visits) and Practice B (1,271 patients; 3,044 visits), we compared regularized logistic regression, random forest, and XGBoost models. Predictors included visit context, prior utilization, and patient characteristics. Models were trained on Practice A and validated on Practice B. ResultsMissed-next-visit rates were 16.2%(A) and 20.7%(B). In external validation, performance was similar across models (AUC 0.66-0.68). At the threshold maximizing F1 score, recall ranged from 0.54-0.71. The LASSO logistic regression model identified six key predictors: timepoint, visit delay, prior no-shows, schedule lead time, new patient status, and immunization refusal. SHAP values confirmed these process measures as among the most influential features across all models. ConclusionPredicting WCV nonattendance is feasible using routine data. A simple logistic regression model performs comparably to complex algorithms, offering a practical pathway for clinical integration. By identifying at-risk families during a current appointment, this may enable clinicians to provide proactive support to support preventive care before a lapse occurs. ImpactO_LIMissed well-child visits are common, leading to an increasing number of preventable acute care visits, delayed recognition of developmental delays, and missed opportunities to initiate early intervention C_LIO_LIA multimodal approach is needed to support well-child visit attendance C_LIO_LIMachine learning is an emerging tool to predict well-child visit no show rates with implications for future interventions to support families at risk for missing well-child visits and promote positive health outcomes C_LI