Abstract
BACKGROUND: Depression is highly prevalent among nursing students (28.7%-30%). Although previous studies have identified multiple influencing factors, the lack of systematic prioritization hinders targeted intervention in resource-limited contexts. This study employed XGBoost and SHAP values to identify and prioritize key risk factors, thereby establishing a data-driven framework to assist educational administrators in optimizing resource allocation and facilitating early detection and personalized support. METHODS: This multicenter cross-sectional study was conducted from September to December 2024 among nursing students recruited from ten universities in Shandong, Jiangxi, Henan, Hubei, and Sichuan provinces. Data were collected using a structured questionnaire comprising a demographic characteristics form, the Center for Epidemiological Studies Depression Scale (CES-D), and the Social Interaction Anxiety Scale (SIAS). Data cleaning was performed in Excel, and statistical analyses were conducted using SPSS Statistics version 27.0 and Python 3.9. RESULTS: The incidence of depression among nursing students is 28.60%. According to the random forest model, the order of depression predicted by this study from high to low is Sleep Condition, Social anxiety, Mother's Educational Level, Sexual Orientation, Smoking, and Household composition. CONCLUSION: Depression is highly prevalent among nursing students, representing a significant challenge to both student well-being and the future healthcare workforce. This study identified and prioritized key determinants of depression, including poor sleep quality, social anxiety, low maternal education, sexual minority status, smoking, and single-parent family background. These findings can provide a basis for nursing administrators and educators to develop targeted and personalized intervention strategies.