Discovery of factors associated with smartphone addiction among high school adolescents: Using machine learning and...

Nowadays, smartphone addiction (SA) has become a widespread issue; however, research specifically targeting high school adolescents remains limited. To address this gap, this study aimed to identify important predictive factors of SA among high school students using machine learning and to explore the relationships among these factors through network analysis, thereby providing a theoretical foundation for targeted interventions.

A total of 14,036 high school students aged 15-18 years (M = 16.40) were included for model training and internal validation. The Extreme Gradient Boosting (XGBoost) algorithm was applied to evaluate the importance of 20 candidate variables, ranked using SHAP values, from which the top nine predictors were selected. Network analysis was then employed to examine the connections among these predictors.

The XGBoost model achieved the highest predictive performance among tested approaches. Network analysis revealed that anxiety was the most central factor, showing positive associations with interpersonal sensitivity and mood swings.

XGBoost combined with network analysis is an effective approach for identifying and exploring SA-related factors among high school adolescents. The identified predictors provide valuable insights for designing targeted and evidence-based interventions for SA in this population.