aceIdentifying key risk factors of adolescent Internet Gaming Disorder using explainable machine learning and network analysis
Abstract: Background Internet Gaming Disorder (IGD) poses significant psychological and social risks for adolescents. Its development involves complex interactions among early adversity, negative life events, psychological symptoms, and regulatory emotional self-efficacy, yet relative importance of these factors and their interrelationships remain incompletely understood. Objective To identify the key demographic and psychosocial factors associated with adolescent IGD, quantify their relative contributions to IGD risk, an…
Student Counseling Sercvices - Texas A&M by Patrick Creighton. CC BY-SA 4.0 · https://creativecommons.org/licenses/by-sa/4.0
By September 2026, researchers reported using four machine learning algorithms to classify Internet Gaming Disorder risk among 6573 adolescents, selecting a random forest model after comprehensive performance evaluation. They applied SHAP to rank feature contributions and network analysis to map interrelationships among psychosocial factors.
The work matters because it moves beyond single-factor associations to quantify relative importance and connectivity of distress, health adaptation problems, early adversity, and stress for adolescent IGD, offering a basis for early risk identification. What remains uncertain is how these ranked and central factors translate into validated screening or intervention in clinical settings.
- Study developed classification models in 6573 adolescents using four machine learning algorithms and selected random forest as final model after evaluation with ROC-AUC, PR-AUC, Brier score and DCA.
- SHAP ranking identified top contributors including gender, depression symptoms, insomnia, social anxiety, emotional abuse, health adaptation problems, anxiety symptoms, academic stress, punishment experiences, and regulating despondency/distress.
- Network analysis found anxiety, academic stress, and health adaptation as central nodes, with strongest positive edges between health adaptation and punishment experiences, and between depression and anxiety.
Random forest with SHAP and network analysis among 6573 adolescents ranked key contributors to IGD risk and identified central nodes, which may inform early risk identification.
The rundown
Researchers built and evaluated four machine learning classification models in a sample of 6573 adolescents to identify demographic and psychosocial factors associated with IGD, using ROC-AUC, PR-AUC, threshold-based metrics, Brier score and DCA, and selected the random forest model for subsequent analyses.
SHAP analysis provided directional associations and ranked contributions, while network analysis characterized interrelationships, revealing central roles for anxiety, academic stress, and health adaptation and strong links between health adaptation and punishment experiences and between depression and anxiety.
Sources
- Peer-reviewedJournal of Affective Disorders2026-09-13
How should this claim be treated?
The debate