Development and validation of a predictive model for depression in patients with advanced stage of cardiovascular-kidney-metabolic syndrome

Depression is highly prevalent among patients with chronic disease advanced and with poor clinical outcomes. However, effective tools for identifying individuals at risk remain limited. This study aimed to develop and validate a predictive model for depression in patients with advanced stage of cardiovascular-kidney-metabolic (CKM) syndrome. A total of 1072 participants from National Health and Nutrition Examination Survey (NHANES) were included, with 750 assigned to the training set and 322 to the test set. The three external validation sets consist of 164, 249, and 166 individuals. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9). Least Absolute Shrinkage and Selection Operator (LASSO) regression was applied to identify key predictors construct model 1. LASSO regression and followed with multivariate logistic regression used to construct the model 2. Random forest, support vector machines, or decision trees were used to construct the model 3, model 4, or model 5. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and calibration plots. Model 2 demonstrated the best performance among all models, with an AUC of 0.768 in the test set. The final model included sleep disorder age, sex, poverty-income ratio, waist circumference, and gamma-glutamyl transferase as significant predictors of depression. External validation showed consistent predictive accuracy, with AUCs ranging from 0.765 to 0.794 across three independent validation sets.