Health and Socio-Demographic Risk Factors of Childhood Stunting: Assessing the Role of Factor Interactions Through the Development of an AI Predictive Model

Abstract

Stunting is a significant global health problem, especially in developing countries such as Indonesia. This study aims to develop and evaluate an artificial intelligence (AI)-based predictive model to identify the risk of stunting in children using the CatBoost algorithm which is a combination of Weighted Apriori and XGBoost. This model is designed to utilize the advantages of each algorithm in handling data with variable weights to improve prediction accuracy. Feature analysis shows that "Height (cm) & Age (months)" are the main indicators in classifying children's nutritional status. Model evaluation shows high accuracy of 94.85%, precision of 95%, recall of 94.85%, and F1 Score of 94.84%. Kappa Coefficient and Matthews Correlation Coefficient (MCC) reached 93.13% and 93.19%, respectively, while ROC-AUC reached 99.70%. These findings indicate that the CatBoost model can provide highly accurate results in detecting the risk of stunting and offer in-depth insights into risk factors that can improve the effectiveness of health interventions. This study fills the gap in the literature by integrating the Weighted Apriori and XGBoost algorithms, providing a significant contribution to early detection of stunting and supporting government efforts to reduce the prevalence of stunting in Indonesia and other regions.