KMAN Publication Inc. (KMANPUB) تست 3060-6713 4 Serial Number 13 2026 04 01 Machine Learning Classification of Suicidal Ideation in Adolescents with Attention-Deficit/Hyperactivity Disorder: A Random Forest Approach Incorporating Impulsivity, Emotional Dysregulation, and Family Cohesion Machine Learning Classification of Suicidal Ideation in Adolescents with Attention-Deficit/Hyperactivity Disorder: A Random Forest Approach Incorporating Impulsivity, Emotional Dysregulation, and Family Cohesion 1 13 10.61838/kman.prien.5399 EN Journal Article 2025 11 12 <p>The present study aimed to develop and evaluate a Random Forest machine learning model for classifying suicidal ideation among adolescents with Attention-Deficit/Hyperactivity Disorder (ADHD) based on impulsivity, emotional dysregulation, and family cohesion. This cross-sectional predictive modeling study was conducted among 468 Canadian adolescents aged 13–18 years with a confirmed diagnosis of ADHD. Participants were recruited from outpatient psychiatric clinics, mental health centers, and educational counseling services across multiple provinces. Suicidal ideation was assessed using the Suicidal Ideation Questionnaire-Junior (SIQ-JR), impulsivity was measured using the Barratt Impulsiveness Scale-11 (BIS-11), emotional dysregulation was evaluated using the Difficulties in Emotion Regulation Scale (DERS), and family cohesion was assessed using the Family Adaptability and Cohesion Evaluation Scales IV (FACES-IV). Data were analyzed using a Random Forest classification algorithm implemented in Python. The dataset was divided into training and testing subsets using an 80:20 ratio, and 10-fold cross-validation was employed to optimize model performance. Classification accuracy, precision, recall, specificity, F1-score, and area under the receiver operating characteristic curve (AUC-ROC) were used to evaluate predictive performance. Feature importance analyses were conducted to identify the relative contribution of each predictor. The Random Forest model demonstrated excellent classification performance, achieving an accuracy of 88.2%, precision of 86.4%, recall of 90.1%, specificity of 86.7%, F1-score of 88.2%, and an AUC-ROC of .934. Suicidal ideation was positively associated with impulsivity (r = .54, p &lt; .01) and emotional dysregulation (r = .67, p &lt; .01) and negatively associated with family cohesion (r = −.49, p &lt; .01). Feature importance analysis revealed that emotional dysregulation was the strongest predictor of suicidal ideation (35.8%), followed by family cohesion (28.5%) and impulsivity (24.4%). The confusion matrix further indicated high sensitivity and low rates of classification error. The findings demonstrate that machine learning approaches can accurately classify suicidal ideation among adolescents with ADHD and highlight emotional dysregulation, family cohesion, and impulsivity as key determinants of suicide risk. These results underscore the importance of integrating psychological and family-related factors into suicide prevention efforts and support the potential clinical utility of machine learning-based screening tools for early identification of high-risk adolescents.</p> https://journals.kmanpub.com/index.php/prien/article/download/5399/10703