Predictors of smartphone addiction in adolescents with depression: combing the machine learning and moderated...

Smartphone addiction (SA) significantly impacts the physical and mental health of adolescents, and can further exacerbate existing mental health issues in those with depression. However, fewer studies have focused on the predictors of SA in adolescents with depression. This study employs machine learning methods to identify key risk factors for SA, using the interpretable SHapley Additive exPlanations (SHAP) method to enhance interpretability. Additionally, by constructing a mediation moderation model, the interactions between significant risk factors are analyzed. The study included 2203 adolescents with depression. Machine learning results from four models (Random Forest, Support Vector Machine, Logistic Regression, XGBoost) consistently identified emotion-focused coping, rumination, and school bullying as the strongest predictors of SA. Further mediation moderation analyses based on the Interaction of Person-Affect-Cognition-Execution (I-PACE) model revealed that rumination significantly mediated the relationship between school bullying and SA, and emotion-focused coping significantly moderated the relationships between school bullying and both rumination and SA. This is the first study to use machine learning to explore the predictors of SA in depressive adolescents and further analyze the interactions among these predictors. Future interventions for SA in adolescents with depression may benefit from psychotherapy that addresses emotion-focused coping and rumination.