KMAN Publication Inc. (KMANPUB) Journal of Adolescent and Youth Psychological Studies (JAYPS) 2981-2526 7 Serial Number 45 2026 05 01 Ensemble Learning Approaches to Predicting Youth Suicidal Ideation Using Emotional Numbing, Cyberbullying Exposure, AI Chatbot Attachment, Hopelessness, and Social Withdrawal Ensemble Learning Approaches to Predicting Youth Suicidal Ideation Using Emotional Numbing, Cyberbullying Exposure, AI Chatbot Attachment, Hopelessness, and Social Withdrawal 1 14 10.61838/kman.jayps.5429 EN Journal Article 2025 09 16 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to investigate the predictive role of emotional numbing, cyberbullying exposure, AI chatbot attachment, hopelessness, and social withdrawal in youth suicidal ideation using advanced ensemble learning approaches among Canadian adolescents and emerging adults.</p> <p><strong>Methods and Materials: </strong>This cross-sectional predictive study was conducted among 742 adolescents and emerging adults aged 16 to 24 years recruited from educational institutions and youth communities across Canada. Data were collected using standardized instruments assessing suicidal ideation, emotional numbing, cyberbullying exposure, hopelessness, social withdrawal, and attachment to AI chatbot systems. After preprocessing procedures including missing data imputation and feature normalization, several ensemble machine learning algorithms including Random Forest, Gradient Boosting Machine, AdaBoost, XGBoost, LightGBM, and a stacked ensemble classifier were implemented using Python-based analytical frameworks. Model performance was evaluated through accuracy, precision, recall, F1-score, specificity, and area under the receiver operating characteristic curve (AUC-ROC). Feature importance and SHAP analyses were additionally performed to examine the relative contribution of predictors within the final classification model.</p> <p><strong>Findings: </strong>The findings demonstrated significant positive relationships between suicidal ideation and all predictor variables, including emotional numbing, cyberbullying exposure, AI chatbot attachment, hopelessness, and social withdrawal (p &lt; .01). Hopelessness emerged as the strongest predictor within the final ensemble model, followed by emotional numbing and social withdrawal. The stacked ensemble classifier demonstrated the highest predictive performance with an accuracy of 94%, precision of 92%, recall of 91%, F1-score of 91%, specificity of 95%, and an AUC-ROC value of 0.97. XGBoost and LightGBM also demonstrated strong classification capability. SHAP analysis confirmed that higher levels of hopelessness, emotional numbing, cyberbullying exposure, social withdrawal, and AI chatbot attachment significantly increased the probability of high-risk suicidal ideation classification.</p> <p><strong>Conclusion: </strong>The findings suggest that youth suicidal ideation is shaped by multidimensional interactions among emotional dysregulation, interpersonal isolation, cybervictimization, hopelessness, and emerging forms of technological attachment. Ensemble learning approaches demonstrated exceptional effectiveness in identifying adolescents at elevated suicide risk and may provide valuable computational tools for early psychological screening and suicide prevention programs. The study further highlights the growing psychological significance of AI chatbot attachment within contemporary youth mental health contexts.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/jayps/article/download/5429/9863