KMAN Publication Inc. (KMANPUB) Psychology of Woman Journal 3041-8615 7 Serial Number 27 2026 05 01 TabNet Prediction of Depression Risk Among Women Through Rumination, Social Isolation, Self-Criticism, and Intolerance of Uncertainty TabNet Prediction of Depression Risk Among Women Through Rumination, Social Isolation, Self-Criticism, and Intolerance of Uncertainty 1 13 10.61838/kman.pwj.5474 EN Journal Article 2026 01 14 <table> <tbody> <tr> <td> <p><strong>Objective:</strong> The present study aimed to predict depression risk among women using rumination, social isolation, self-criticism, and intolerance of uncertainty through an explainable TabNet machine learning model and to determine the relative importance of these predictors in identifying individuals at elevated risk for depressive symptoms.</p> <p><strong>Methods and Materials: </strong>This cross-sectional predictive study was conducted among 1,248 women residing in Kenya. Participants were recruited using a stratified convenience sampling strategy from community health centers, educational institutions, workplaces, and online platforms. Data were collected using standardized self-report measures, including the Patient Health Questionnaire-9 (PHQ-9) for depression risk, the Ruminative Responses Scale (RRS), the Lubben Social Network Scale-Revised (LSNS-6), the Levels of Self-Criticism Scale (LOSC), and the Intolerance of Uncertainty Scale-12 (IUS-12). Following data preprocessing and standardization, the dataset was divided into training (70%), validation (15%), and testing (15%) subsets. A TabNet deep learning model was developed and optimized using cross-validation and hyperparameter tuning procedures. Model performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean squared error (MSE). Feature importance and SHAP analyses were conducted to enhance interpretability.</p> <p><strong>Findings: </strong>The TabNet model demonstrated excellent predictive performance, achieving an R² of .874 in the training set, .846 in the validation set, and .832 in the testing set. Correlation analyses indicated significant positive associations between depression risk and rumination (r = .71), self-criticism (r = .68), intolerance of uncertainty (r = .65), and social isolation (r = .63). Feature importance analysis revealed that rumination was the strongest predictor (33.8%), followed by self-criticism (27.9%), intolerance of uncertainty (22.1%), and social isolation (16.2%). SHAP analysis confirmed that higher levels of all four predictors contributed to increased depression risk. Classification analyses further demonstrated strong model accuracy, yielding an overall F1-score of .86 and particularly high performance in identifying severe depression risk cases.</p> <p><strong>Conclusion: </strong>The findings indicate that depression risk among women can be accurately predicted using a combination of cognitive and social vulnerability factors. Rumination emerged as the most influential predictor, highlighting the central role of repetitive negative thinking in depressive symptomatology. The results support transdiagnostic models of emotional disorders and demonstrate the utility of explainable machine learning approaches for mental health screening, early detection, and personalized intervention planning.</p> </td> </tr> </tbody> </table> https://journals.kmanpub.com/index.php/psywoman/article/download/5474/10737