Machine learning-driven identification of key risk factors for predicting depression among nurses

Area of Science:

• Medical Informatics
• Public Health
• Psychiatry

Background:

• The COVID-19 pandemic, caused by SARS-CoV-2, has posed a significant global health threat.
• Relaxation of pandemic control policies in China in 2022 increased infection risks for nursing personnel.
• Nursing staff faced heightened risks and potential mental health challenges during the pandemic.

Purpose of the Study:

• To identify risk factors for depression among nursing staff in China following the full reopening of COVID-19 measures in 2022.
• To develop and validate a predictive model for assessing depression risk in this population.
• To provide tools for early identification and intervention for at-risk nursing staff.

Main Methods:

• A cross-sectional study involving 293 nursing staff was conducted in Anhui Province, China.
• Machine learning models including logistic regression, support vector machine (SVM), extreme gradient boosting machine (XGBoost), and adaptive boosting (AdaBoost) were developed.
• Models were trained and validated using 10-fold cross-validation, with performance evaluated by the area under the receiver operating characteristic curve (AUC).

Main Results:

• The XGBoost model achieved the highest predictive accuracy with an AUC of 0.95 and an F1 score of 0.90.
• Logistic regression, SVM, and AdaBoost models also demonstrated strong predictive capabilities.
• The developed models effectively identified common risk factors for depression among Chinese nursing staff.

Conclusions:

• The XGBoost model offers a robust tool for predicting depression risk in nursing staff, aiding clinical managers in timely interventions.
• Findings highlight the need for tailored mental health support for healthcare workers, considering varying global work environments.
• Future research should focus on larger, multi-center studies to validate the model and explore additional risk factors for nursing staff mental well-being.