Development and temporal validation of a machine learning-based prediction model for long-term depressive symptoms in older adults with cardiovascular disease or hypertension

Abstract

Older adults with cardiovascular disease or hypertension have an elevated risk of depressive symptoms, which may adversely affect prognosis and quality of life. A practical prediction model could support early risk stratification and targeted screening. Data were obtained from the China Health and Retirement Longitudinal Study from 2011 to 2020. A time-based design was used for model development and temporal validation. The development cohort included 548 participants followed from 2011 to 2015, and the nonoverlapping temporal validation cohort included 523 participants followed from 2015 to 2020. The Boruta algorithm was used for predictor selection, 7 machine learning algorithms were compared, and Shapley additive explanations were used to interpret the selected model. Missing values were imputed using a random forest algorithm. Seven predictors were selected. Logistic regression showed the best overall performance, with an area under the receiver operating characteristic curve of 0.844 (95% confidence interval, 0.773–0.915) in the development cohort and 0.840 (95% confidence interval, 0.799–0.881) in the temporal validation cohort. The logistic regression model demonstrated good discrimination and calibration for predicting long-term depressive symptoms among older adults with cardiovascular disease or hypertension. Further independent and prospective validation is required before routine clinical implementation.