Abstract
Depression is a significant issue among cancer survivors, and timely identification of depressive symptoms is crucial. This study aims to develop and evaluate a predictive model for depression risk in cancer survivors. 2279 cancer survivors in the National Health and Nutrition Examination Survey were included. Participants were randomly allocated to training and validation sets in a 7:3 ratio. Least absolute shrinkage and selection operator and multivariate logistic regression identified independent predictors of depression (defined as PHQ-9 ≥ 10), which were used to develop a nomogram. Model performance was assessed using receiver operator characteristic, the calibration curve, the Hosmer-Lemeshow test, and the decision curve analysis. Seven variables were identified as significant predictors for depression in cancer survivors: age, education level, poverty-to-income ratio, smoking status, congestive heart failure, sleep disorders, and number of cancers. A nomogram was developed using the 7 predictors. The area under the curve for the model's training and validation sets was 0.802 (95% confidence interval [CI]: 0.767-0.836) and 0.794 (95% CI: 0.740-0.849), respectively. Internal validation via bootstrapping yielded an optimism-corrected area under the curve of 0.812 (95% CI: 0.784-0.840). Calibration curves and the Hosmer-Lemeshow test illustrated the model has favorable calibration capability. Decision curve analysis results demonstrated that the model has satisfactory clinical application. This study developed a nomogram to predict depression risk in cancer survivors, demonstrating potential clinical utility for identifying high-risk individuals.