Digital Vital Signs: Decision Trees as Behavioral Tripwires for Adolescent Smartphone Overuse

Smartphones are a double-edged sword for teenagers; on the one hand, these devices provide a window to vast knowledge. However, the dark side of smartphones emerges when uncontrolled use is linked to mental health and exposure to negative content. Problematic smartphone use (PSU) occurs in 12–37% of adolescents and has been associated with sleep disturbances, depressive symptoms, and deterioration in academic functioning. Methods: We have trained an interpretable decision tree over a 1,000-participant dataset using stratified 80:20 splitting, class balancing, one-hot encoding, and grid search using cross-validation. Results: The model achieved 85.2% test accuracy (CV mean 85.0% ± 1.5%). Primary predictors were screen time per day (risk for >5.3 h/day associated with 4.3× increased risk), social media exposure (more than >2 h/day), and app variety (more than >5 apps/day). Extractable rules (e.g., >6.5 h screen time ∧ >2 h social media 92% precision for "high" addiction) permit tiered intervention thresholds. Conclusions: An interpretable decision tree provides strong prediction and converts insights into actionable behavioral thresholds for parents, schools, and developers for the purpose of early PSU intervention.