Analyzing the Impact of Screen Time on Child and Adolescent Behavior and Academic Performance using Machine Learning Models

Excessive screen time has become a global concern with potential implications on the behavioral and academic outcomes of youth. In this study, we analyze a dataset of children aged 5–15 to explore how educational and recreational screen time relate to academic performance and behavior. We engineer three synthetic features - Academic Performance Score, Behavioral Score, and Overall Impact- based on rule-based thresholds (educational use ≤2 hours, recreational use <2 hours). We then apply machine learning models (Decision Tree, Random Forest, Logistic Regression, Support Vector Machine, XGBoost) to predict overall impact and compare their performance. The Random Forest model achieved the best accuracy (~96%). Our findings indicate that higher recreational screen time (especially beyond 2 hours daily) is associated with poorer behavioral outcomes and lower academic performance, particularly in older children. We discuss these results in the context of existing literature and provide recommendations for parents and policymakers to promote balanced screen use. Future work should consider more diverse and longitudinal datasets to further validate these insights.