Predictive Modeling of Psychiatric Appointment No-Shows Using Multidomain Factors

We analyzed 32,568 psychiatric appointments (5,000 patients, 7.8% no-show rate) integrating weather, economic, and geographic factors. Gas prices, precipitation, and early appointments significantly predicted no-shows (p < .001). XGBoost achieved 91% accuracy (AUC-ROC = 0.897). Ensemble modeling approaches optimize sensitivity-specificity tradeoffs for targeted clinical interventions.