AI Summary & Key Takeaways
AI InsightThe results emphasize the need for school- and family-based interventions to reduce digital addiction and prevent digital obesity in adolescents.
📌 3 Key Takeaways
Abstract
This study aimed to determine the levels of problematic internet use, digital gaming addiction, and digital obesity among high school students, to examine their interrelationships, and to predict digital obesity using machine learning methods. Between June 2024 and June 2025, a descriptive, correlational, and predictive modelling design was employed in four public high schools in XXX. The study sample consisted of 490 students selected through a multistage sampling process. Data were collected using a personal information form, the Problematic Internet Use Scale, and the Digital Game Addiction Scale. For the prediction of digital obesity, feature selection was performed with the Boruta algorithm and seven machine learning models. Problematic internet use and digital gaming addiction were at moderate levels, both significantly higher among male and low-income students (p < 0.05). A moderate positive correlation was found between the two behavioral variables. Among machine learning models, random forest and support vector machine achieved the highest predictive accuracy for digital obesity, with weekend screen time emerging as the most important predictor. Machine learning methods demonstrated high accuracy in predicting digital obesity and revealed strong associations between digital dependency behaviors. The results emphasize the need for school- and family-based interventions to reduce digital addiction and prevent digital obesity in adolescents.