Development and validation of a risk prediction model for depression in community-based chronic obstructive pulmonary disease patients using machine learning techniques

Abstract

Background: Chronic obstructive pulmonary disease (COPD) is a common respiratory condition that presents a significant health challenge. Research suggests that individuals with COPD frequently experience symptoms of depression. Despite this association, there is a marked lack of reliable predictive models that can accurately forecast the likelihood of depression in community-dwelling COPD patients. This study aims to develop an innovative predictive model utilizing machine learning techniques to effectively anticipate the risk of depression in individuals with COPD living in the community. Methods: The research utilized data from the 2020 China Health and Retirement Longitudinal Study (CHARLS) database to examine clinical information from a sample of 809 COPD patients residing in the community. The analysis involved the application of the Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression to identify predictive factors. Furthermore, various machine learning (ML) classification models were employed, with the most optimal model being determined and assessed. Additionally, a personalized risk assessment tool using Shapley Additive exPlanations (SHAP) was developed for interpreting the findings. Result: The study identified six key indicators of depression in community-based COPD patients. These included Self_assessed_health, Sleeping_time_at_night, Memory_disease, Gender, ADL_ score, and Age. Logistic classification model was the optimal model, test set area under curve (AUC) (95% confidence interval, CI):0.713 (0.648-0.778). Conclusions:The model constructed in this study has relatively reliable predictive performance, which helps clinical doctors identify high-risk populations of community COPD patients prone to depression at an early stage.