Previously submitted to: JMIR AI (no longer under consideration since Mar 26, 2026)
Date Submitted: Apr 7, 2025
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Development of an Interpretable Machine Learning Model to Predict Burnout and Protective Factors Among Healthcare Professionals: A Cross-Sectional Study Using XGBoost, SHAP, and BATEREP Data
ABSTRACT
Background:
Burnout syndrome is a critical occupational health concern among healthcare professionals, characterized by chronic emotional exhaustion, depersonalization, and reduced personal accomplishment. Sustained exposure to high-stress environments, excessive workloads, administrative burdens, and poor work-life balance contributes to adverse health outcomes and compromised clinical performance.
Objective:
This study aimed to develop and evaluate an interpretable machine learning model for predicting burnout severity and identifying protective factors among healthcare professionals, using a large-scale dataset and validated clinical, psychological, and occupational indicators.
Methods:
We conducted a cross-sectional study using the BATEREP dataset, which included 16,000 anonymized responses from healthcare professionals. Variables encompassed physiological stress markers (e.g., heart rate variability, blood pressure, sleep disturbances), psychological parameters (e.g., depression and anxiety scores, chronic fatigue), cognitive performance indicators (e.g., working memory, executive dysfunction), and occupational factors (e.g., patient load, night shifts, administrative burden, perceived work-life balance). We trained a machine learning model using XGBoost and applied SHAP (SHapley Additive exPlanations) to interpret feature contributions and identify key predictors of burnout risk.
Results:
The XGBoost model achieved high predictive performance (AUC = 0.87) in identifying individuals at elevated burnout risk. SHAP analysis revealed that emotional exhaustion, perceived work-life imbalance, high psychological distress scores, and impaired cognitive performance were among the strongest contributors to burnout prediction. Conversely, adequate sleep quality, manageable workloads, and supportive work environments emerged as significant protective factors.
Conclusions:
Our findings demonstrate the utility of interpretable machine learning models in identifying burnout risk and protective factors among healthcare professionals. Integrating such tools into occupational health screening protocols may enable early intervention and targeted support strategies. Future research should explore the implementation of predictive models in real-world clinical settings and evaluate their impact on healthcare workforce well-being and performance.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.