건강검진서비스의 예약 부도(No-Show) 후 재예약 및 방문 여부 예측

초록

Purpose: No-shows for health screening appointments place significant operational strain on healthcare providers, underscoring the need for effective management and response strategies. This study seeks to enhance the predictive accuracy of customer rebooking and attendance behaviors by analyzing 1,010,908 no-show records collected by the Korean Association of Health Promotion between January 1, 2020, and December 31, 2022. Methodology: A variety of machine learning and deep learning models were applied to predict rebooking and attendance behaviors, including Decision Trees, Random Forest, K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Naive Bayes, Long Short-Term Memory (LSTM), CatBoost, XGBoost, and LightGBM. Model performance was assessed through confusion matrices, categorical crossentropy to evaluate accuracy, precision, and recall. Findings: The LightGBM model outperformed other models, achieving the highest accuracy (0.8649), precision (0.6708), and F1 score (0.4998), followed by Catboost and XGBoost. Practical Implications: The results of this study provide healthcare providers with a deeper understanding of the behavioral patterns of clients who fail to attend scheduled appointments. These insights can support the development of more effective strategies to encourage rebooking, enhance operational efficiency, and reduce losses associated with no-shows.

키워드