KMAN Publication Inc. (KMANPUB) تست 3060-6713 In Press 2025 12 01 A LightGBM-Based Prediction Model of Emotional Dysregulation in Children with ADHD: The Roles of Parent–Child Conflict, Cognitive Control, and Sleep Quality A LightGBM-Based Prediction Model of Emotional Dysregulation in Children with ADHD: The Roles of Parent–Child Conflict, Cognitive Control, and Sleep Quality 1 13 EN Journal Article 2026 04 14 <table> <tbody> <tr> <td> <p>The present study aimed to develop and evaluate a Light Gradient Boosting Machine (LightGBM) model for predicting emotional dysregulation in children with Attention-Deficit/Hyperactivity Disorder (ADHD) based on parent–child conflict, cognitive control difficulties, and sleep quality, while identifying the relative contribution of each predictor using explainable artificial intelligence techniques. This cross-sectional predictive modeling study was conducted among 428 Canadian children diagnosed with ADHD between the ages of 8 and 12 years. Participants were recruited from pediatric neuropsychology clinics, child mental health centers, and educational support programs. Emotional dysregulation was assessed using the Emotion Regulation Checklist, parent–child conflict was measured using the Child–Parent Relationship Scale, cognitive control difficulties were evaluated through the Behavior Rating Inventory of Executive Function–Second Edition, and sleep quality was assessed using the Children’s Sleep Habits Questionnaire. Data preprocessing included standardization, missing-value imputation, and outlier screening. The dataset was divided into training (80%) and testing (20%) subsets. A LightGBM algorithm with five-fold cross-validation and Bayesian hyperparameter optimization was employed. Model performance was evaluated using R², RMSE, MAE, and MAPE indices, and predictor importance was examined using Shapley Additive Explanations (SHAP). Correlation analyses indicated significant positive associations between emotional dysregulation and parent–child conflict (r = .68, p &lt; .001), cognitive control difficulties (r = .61, p &lt; .001), and poor sleep quality (r = .57, p &lt; .001). The optimized LightGBM model demonstrated excellent predictive performance, explaining 84.2% of variance in the training dataset and 78.1% of variance in the testing dataset. Cross-validation yielded a mean R² of .792, indicating strong model stability. SHAP analyses revealed that parent–child conflict was the most influential predictor (41.8%), followed by cognitive control difficulties (33.6%) and poor sleep quality (24.6%). Higher levels of each predictor were associated with increased emotional dysregulation. The findings demonstrate that emotional dysregulation in children with ADHD can be accurately predicted using a machine learning framework that integrates family, cognitive, and sleep-related factors. Parent–child conflict emerged as the strongest determinant, highlighting the importance of family relationships in emotional functioning. The results support the application of explainable artificial intelligence approaches for identifying clinically meaningful risk factors and informing personalized assessment and intervention strategies for children with ADHD.</p> </td> </tr> </tbody> </table> https://www.journals.kmanpub.com/index.php/prien/article/download/5414/12137