Research on the Prediction Model of Occupational burnout Among Primary Healthcare Workers Based on Machine Learning Algorithms

Objective: To conduct a questionnaire survey among primary-level medical personnel in Jinan City, Shandong Province, China, and to construct and evaluate machine learning models in predicting job burnout among this group. Methods: A stratified cluster sampling method was adopted in this study to select 1,710 primary-level medical personnel in Jinan City, Shandong Province, as study participants. A general information questionnaire, the Minnesota Satisfaction Questionnaire - Short Form (MSQ-SF), and the Maslach Burnout Inventory (MBI) were utilized to investigate demographic data, job satisfaction, and job burnout. Three machine learning algorithms—Support Vector Machine, Logistic Regression, and Decision Tree—were employed to construct prediction models. Model performance was evaluated, and predictive feature importance was analyzed. Results: The incidence of job burnout among primary-level medical personnel in Jinan City was relatively high. The Support Vector Machine model demonstrated superior performance, with job satisfaction identified as the most significant predictor. Conclusion: It is recommended to fully harness the practical value of the job burnout prediction model for primary-level medical personnel. Early identification of burnout should be prioritized, and targeted measures should be implemented to alleviate burnout and reduce its negative impacts.