Adolescent Depression Risk Prediction Using Machine Learning and Behavioral Health Data Integration

Abstract

Background: Depression is one of the most prevalent mental health disorders among adolescents and is associated with significant emotional, social, and academic consequences. Early identification of depression risk is essential for timely intervention and prevention. Advances in machine learning (ML) techniques and behavioral health data integration provide new opportunities for predicting depression risk with improved accuracy. By analyzing behavioral, psychological, social, and lifestyle indicators, ML models can assist healthcare professionals in identifying vulnerable adolescents before severe symptoms develop.

Objective: This study aimed to develop and evaluate a machine learning-based framework for predicting adolescent depression risk using integrated behavioral health data.

Methodology: A dataset comprising 500 adolescents aged 13–19 years was analyzed. Behavioral health variables, including anxiety scores, sleep quality, physical activity, social support, academic stress, and digital behavior indicators, were integrated into a machine learning model. Data were preprocessed and analyzed using Random Forest, Support Vector Machine (SVM), and Logistic Regression algorithms. Model performance was assessed using accuracy, precision, recall, and F1-score metrics.

Findings: The Random Forest model achieved the highest predictive performance with 91.4% accuracy, 89.8% precision, 90.6% recall, and an F1-score of 90.2%. Academic stress, sleep quality, anxiety levels, and social support emerged as the most influential predictors of depression risk.

Conclusion: Machine learning combined with behavioral health data integration demonstrates strong potential for accurately predicting adolescent depression risk. Such predictive systems may support early detection, targeted interventions, and improved mental health outcomes among adolescents.