Explainable XGBoost Models for Predicting Burnout among Mental Health Professionals Using Compassion Fatigue, Psychological Capital, Self-Care, and Occupational Stress
Journal of Assessment and Research in Applied Counseling · 2026 · 8 (3) · 1-15
DOI: 10.61838/kman.jarac.5469 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Assessment and Research in Applied Counseling, 2026. Verfasst von Abdul-Wahab Inusah und 3 weiteren Autor:innen.
Themen: Burnout in der Pflege, Selbstmanagement bei chronischen Erkrankungen
Abstract (Originalsprache)
Objective
This study aimed to develop and interpret an explainable XGBoost model for predicting burnout among mental health professionals in Malaysia using compassion fatigue, psychological capital, self-care, and occupational stress as primary predictors. Methods and Materials: This cross-sectional predictive study was conducted among 742 mental health professionals in Malaysia, including counselors, clinical psychologists, psychiatric nurses, psychiatrists, social workers, psychotherapists, and other mental health specialists. Burnout was assessed using the Maslach Burnout Inventory–Human Services Survey, compassion fatigue using the Professional Quality of Life Scale, psychological capital using the Psychological Capital Questionnaire, self-care using the Self-Care Assessment Worksheet, and occupational stress using the Health and Safety Executive Management Standards Indicator Tool. Data were analyzed using Python-based machine learning procedures. After preprocessing, the dataset was divided into training and testing sets using stratified sampling. An XGBoost classifier was trained and optimized through five-fold cross-validation and hyperparameter tuning. Model performance was evaluated using accuracy, precision, recall, specificity, F1-score, Matthews correlation coefficient, AUC-ROC, and Brier score. SHAP analysis was applied to interpret global and individual predictor contributions.
Findings
Compassion fatigue showed a strong positive association with burnout, while occupational stress also demonstrated a strong positive relationship with burnout. Psychological capital and self-care were negatively associated with burnout, indicating their protective role. The optimized XGBoost model demonstrated excellent predictive performance, with test accuracy of 91.9%, precision of 91.3%, recall of 90.7%, specificity of 92.6%, F1-score of 91.0%, Matthews correlation coefficient of 0.864, AUC-ROC of 0.962, and Brier score of 0.071. SHAP analysis identified compassion fatigue as the most influential predictor, followed by occupational stress, psychological capital, and self-care.
Conclusion
Explainable XGBoost modeling provided an accurate and interpretable approach for predicting burnout among Malaysian mental health professionals. The findings emphasize that burnout is primarily shaped by modifiable psychological and occupational factors, particularly compassion fatigue, occupational stress, psychological capital, and self-care.
Zitieren
APA
Inusah, A., Yang, T., Washim, M. R., & Mustapa, M. (2026). Explainable XGBoost Models for Predicting Burnout among Mental Health Professionals Using Compassion Fatigue, Psychological Capital, Self-Care, and Occupational Stress. Journal of Assessment and Research in Applied Counseling, 8(3), 1-15. https://doi.org/10.61838/kman.jarac.5469
Vancouver
Inusah A, Yang T, Washim MR, Mustapa M. Explainable XGBoost Models for Predicting Burnout among Mental Health Professionals Using Compassion Fatigue, Psychological Capital, Self-Care, and Occupational Stress. Journal of Assessment and Research in Applied Counseling. 2026;8(3):1-15. doi:10.61838/kman.jarac.5469
BibTeX
@article{Inusah2026,
author = {Inusah, Abdul-Wahab and Yang, Ting and Washim, Mohammad Rahman and Mustapa, Mohamad},
title = {Explainable XGBoost Models for Predicting Burnout among Mental Health Professionals Using Compassion Fatigue, Psychological Capital, Self-Care, and Occupational Stress},
journal = {Journal of Assessment and Research in Applied Counseling},
year = {2026},
volume = {8},
number = {3},
pages = {1--15},
doi = {10.61838/kman.jarac.5469}
}
EUVIMED