How to Predict the Onset of Anxiety and Depression in Pediatric Residents? -- A Survey From China.

Purpose: Pediatric residents are a high-risk group, prone to negative emotions such as anxiety and depression. Previous studies have explored a variety of factors affecting the mental health of medical staff, but few have utilized machine learning methods for early warning prediction of psychological problems. This study aimed to investigate the mental health status of pediatric residents in two Chinese cities, Chongqing and Hangzhou, and develop a predictive model for identifying anxiety and depression. Methods: The model was validated using data from both cities and found to be robust. The best predictor variables included sociodemographic information, emotion self-rating, social support rating, and psychological resilience. The study utilized 18 mainstream machine learning methods and screened the best early warning prediction model using the cross-validated LASSO method. Results: The Linear Discriminant Analysis (LDA) model had the best discriminatory performance in distinguishing the presence or absence of anxiety and depression in pediatric residents. The tuned LDA model exhibits excellent discriminatory performance (AUC = 0.923) on the test set. Stress and psychological resilience, particularly commiseration, were found to be strong predictors of anxiety and depression. Conclusion: The study provides a new perspective on early warning prediction and intervention for negative emotions in pediatric residents and medical staff, which may be a new direction for interventions to address mental health issues in this group.