KMAN Publication Inc. (KMANPUB) Applied Family Therapy Journal (AFTJ) 3041-8798 6 Serial Number 30 2025 11 01 Deep Learning Analysis of Family Communication Networks and Adolescent Risk Behavior Prediction Deep Learning Analysis of Family Communication Networks and Adolescent Risk Behavior Prediction 1 10 10.61838/kman.aftj.4949 EN Journal Article 2024 05 24 Objective: The objective of this study was to develop and evaluate a deep learning model capable of predicting adolescent risk behavior based on the structural and relational characteristics of family communication networks. Methods and Materials: This applied, predictive study employed a mixed-methods design with a dominant quantitative approach and was conducted among 782 high school students aged 14–18 years in Georgia, United States. Multi-stage cluster sampling was used to ensure demographic representation. Data were collected using standardized instruments assessing family communication quality, parental monitoring, emotional expressiveness, conflict resolution, and adolescent risk behaviors. Family interactions were transformed into weighted communication networks and analyzed using a hybrid deep learning architecture integrating graph neural networks and long short-term memory models. Model performance was evaluated using accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve, with comparative analysis against traditional machine learning models. Findings: The deep learning model achieved high predictive performance (test accuracy = 0.87; AUC = 0.90), significantly outperforming random forest, support vector machine, and logistic regression models. Communication openness, network density, and parental monitoring emerged as the strongest predictors of adolescent risk behavior. Explainable AI analysis confirmed that fragmented communication networks and negative emotional tone substantially increased predicted risk levels. Conclusion: The findings demonstrate that adolescent risk behavior can be accurately predicted through deep learning analysis of family communication networks, highlighting the critical importance of relational structure and communication quality in prevention and early intervention strategies.

https://journals.kmanpub.com/index.php/aftj/article/download/4949/8908