Development, validation, and visualization of a novel nomogram to predict depression risk in patients with stroke.

Background: This study aimed to develop and validate a predictive nomogram model applicable to depression risk in stroke patients.

Methods: Participants from the NHANES database (n = 1097) were enrolled from 2005 to 2018; 767 subjects were randomly assigned to the training cohort, and the remaining subjects composed the testing cohort. A nomogram containing the optimal predictors identified by the least absolute shrinkage and selection operator (LASSO) and logistic regression methods was constructed to estimate the probability of depression in stroke patients. To evaluate the performance of the nomogram, the area under the receiver operating characteristic curve (AUC), calibration plot, decision curve analysis (DCA) and internal validation were utilized.

Results: Age, family income, trouble sleeping, coronary heart disease, and total cholesterol were included in the nomogram after filtering predictive variables. The AUCs of the nomogram for the training and testing cohorts were 0.782 (95 % CI = 0.742-0.821) and 0.755 (95 % CI = 0.675-0.834), respectively. The calibration plot revealed that the predicted probability was extremely close to the actual probability of depression occurrence in both the training and testing cohorts. DCA revealed that the nomogram model in the training and testing cohorts had a net benefit when the risk thresholds were 0-0.59 and 0-0.375, respectively.

Limitations: This study was limited by the absence of clinical external validation, which hindered the estimation of the nomogram's external applicability. In addition, this study has a cross-sectional design.

Conclusions: A novel nomogram was successfully constructed and proven to be beneficial for identifying individuals at high risk for depression among stroke patients.