Developing a machine learning-based predictive model for depression risk in patients with cardiovascular diseases.

BACKGROUND: Cardiovascular Diseases (CVD) are frequently comorbid with depression, significantly affecting patient prognosis and quality of life. This study aimed to develop and validate a prediction model using Machine Learning (ML) for estimating current depression risk in patients with CVD. METHODS: Data were obtained from the China Health and Retirement Longitudinal Study (CHARLS). The 2020 wave was used for model training and testing, while the 2018 wave served as temporal validation. Eight ML models, including Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Gradient Boosting Tree (GBT), Adaptive Boosting (AdaBoost), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN), were employed to construct a depression risk prediction model for patients with CVD. The model performance was evaluated using the receiver operating characteristic (ROC) curve, precision-recall (PR) curve, calibration curve, and decision curve analysis (DCA), while model interpretability was assessed via the Shapley additive explanations (SHAP) method. RESULTS: The AdaBoost model showed superior performance. SHAP analysis revealed life satisfaction, instrumental activities of daily living (IADL), sleep time, and self-rated health as the top four predictors of depression risk. CONCLUSION: We developed a model for estimating current depression risk in patients with CVD, which may aid in the early identification of high-risk individuals.