KMAN Publication Inc. (KMANPUB) Journal of Adolescent and Youth Psychological Studies (JAYPS) 2981-2526 7 Serial Number 46 2026 06 10 XGBoost Prediction of Eating Disorder Symptom Severity among Adolescent Girls Using Body Dissatisfaction, Social Comparison, Self-Esteem, and Instagram Use Patterns XGBoost Prediction of Eating Disorder Symptom Severity among Adolescent Girls Using Body Dissatisfaction, Social Comparison, Self-Esteem, and Instagram Use Patterns 1 13 10.61838/kman.jayps.5440 EN Journal Article 2025 12 18 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to examine the relative contributions of body dissatisfaction, social comparison, self-esteem, and Instagram use patterns in predicting eating disorder symptom severity among adolescent girls using the Extreme Gradient Boosting (XGBoost) machine learning algorithm.</p> <p><strong>Methods and Materials: </strong>This cross-sectional predictive study was conducted among 1,284 adolescent girls aged 13–18 years recruited from secondary schools across Canada. Participants completed a battery of validated self-report measures assessing eating disorder symptom severity, body dissatisfaction, social comparison orientation, self-esteem, and Instagram-related behaviors. Eating disorder symptoms were measured using the Eating Disorder Examination Questionnaire, body dissatisfaction was assessed using the Body Shape Questionnaire, social comparison was evaluated through the Iowa-Netherlands Comparison Orientation Measure, and self-esteem was measured using the Rosenberg Self-Esteem Scale. Instagram use patterns included daily usage duration, appearance-focused content viewing, influencer engagement, photo-editing behaviors, and emotional sensitivity to social feedback. Data preprocessing procedures included missing-value treatment, feature normalization, and quality screening. The dataset was divided into training and testing subsets using an 80:20 ratio. An optimized XGBoost regression model was developed using five-fold cross-validation and hyperparameter tuning. Model performance was evaluated using the coefficient of determination (R²), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). SHapley Additive exPlanations (SHAP) analyses were employed to interpret feature importance and predictor contributions.</p> <p><strong>Findings: </strong>The XGBoost model demonstrated strong predictive performance, explaining 81.6% of the variance in eating disorder symptom severity within the testing dataset (R² = 0.816). Body dissatisfaction emerged as the strongest predictor, accounting for the largest proportion of predictive gain, followed by appearance-focused Instagram viewing, self-esteem, and social comparison. SHAP analyses indicated that higher levels of body dissatisfaction, greater engagement with appearance-oriented Instagram content, stronger social comparison tendencies, more frequent photo-editing behaviors, longer Instagram use duration, greater influencer engagement, and heightened emotional sensitivity to likes and comments were associated with increased predicted eating disorder symptom severity. In contrast, higher self-esteem exerted a substantial protective effect and was associated with lower predicted symptom severity. The consistency between training and testing performance metrics indicated minimal overfitting and strong model generalizability.</p> </td> </tr> </tbody> </table> <p><strong>Conclusion: </strong>The findings demonstrate that eating disorder symptom severity among adolescent girls can be predicted with high accuracy through the combined assessment of psychological vulnerabilities and Instagram-related behavioral patterns. Body dissatisfaction appears to represent the central risk factor, while appearance-focused social media engagement, social comparison processes, and diminished self-esteem further amplify vulnerability. These results support integrated theoretical models of eating disorder development and highlight the value of machine learning approaches in identifying high-risk individuals. Prevention and intervention programs should simultaneously target body image concerns, self-esteem enhancement, social comparison reduction, and healthy social media engagement to reduce eating disorder risk among adolescent girls.</p> https://journals.kmanpub.com/index.php/jayps/article/download/5440/10537