KMAN Publication Inc. (KMANPUB) Journal of Assessment and Research in Applied Counseling (JARAC) 8 Serial Number 30 2026 07 01 Predicting Major Depressive Disorder Using Random Forest Models Based on Psychological, Behavioral, and Lifestyle Indicators Predicting Major Depressive Disorder Using Random Forest Models Based on Psychological, Behavioral, and Lifestyle Indicators 1 15 EN Journal Article 2026 03 16 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> This study aimed to develop and evaluate a Random Forest classification model for predicting Major Depressive Disorder among Canadian adults using integrated psychological, behavioral, and lifestyle indicators.</p> <p><strong>Methods and Materials:</strong> This cross-sectional predictive study was conducted among 1,742 adults residing in Canada. Participants completed standardized self-report instruments assessing depressive symptoms, anxiety, stress, emotion regulation difficulties, perceived stress, sleep quality, physical activity, and lifestyle behaviors. Major Depressive Disorder status was determined using the Patient Health Questionnaire-9 cut-off score for clinically significant depressive symptoms. Psychological indicators included DASS depression, anxiety, and stress scores, perceived stress, and emotion regulation difficulties. Behavioral and lifestyle variables included sleep quality, sleep duration, physical activity, screen time, body mass index, alcohol consumption, and demographic characteristics. Data preprocessing included missing-value management, categorical encoding, and feature preparation. The dataset was divided into training and testing subsets using stratified sampling. A Random Forest classification algorithm was trained and optimized through five-fold cross-validation and grid-search hyperparameter tuning. Model performance was evaluated using accuracy, sensitivity, specificity, precision, F1-score, balanced accuracy, ROC-AUC, Cohen’s Kappa, Matthews Correlation Coefficient, and feature-importance analysis.</p> <p><strong>Findings:</strong> The optimized Random Forest model demonstrated strong predictive performance on the independent test dataset, with accuracy of 91.38%, sensitivity of 89.12%, specificity of 92.31%, precision of 87.64%, F1-score of 88.37%, balanced accuracy of 90.72%, ROC-AUC of 0.957, Cohen’s Kappa of 0.804, and Matthews Correlation Coefficient of 0.806. Five-fold cross-validation confirmed model stability, with mean accuracy of 91.38%, mean precision of 87.89%, mean recall of 89.14%, mean F1-score of 88.51%, and mean ROC-AUC of 0.957. Feature-importance analysis identified DASS depression, perceived stress, emotion regulation difficulties, sleep quality, anxiety, stress, screen time, and physical activity as the strongest predictors.</p> <p><strong>Conclusion:</strong> The findings indicate that Random Forest modeling can accurately predict probable Major Depressive Disorder using psychological, behavioral, and lifestyle indicators. The model showed high discrimination, stable validation performance, and clinically interpretable predictor patterns, supporting its potential value as a scalable screening approach for identifying adults at elevated risk of depression.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/jarac/article/download/5450/10923