A predictive model achieved area under the curve (AUC) values of 0.76 in training and 0.73 in validation for identifying depression risks in 3,107 disabled elderly individuals.
Poor self-rated health, pain, lack of caregivers, cognitive impairment, and shorter sleep duration are linked to increased depression risk in this population.
The XGBoost model outperformed other models during training, while logistic regression showed better results during validation.
A good model fit was indicated by the calibration curve and a Brier score of 0.20.
Decision curve analysis supports the model's clinical utility in assessing depression risks.
Simplified
BACKGROUND: Given the accelerated aging population in China, the number of disabled elderly individuals is increasing, and depression is a common mental disorder among older adults. This study aims to establish an effective model for predicting depression risks among disabled elderly individuals.
METHODS: The data for this study was obtained from the 2018 China Health and Retirement Longitudinal Study (CHARLS). In this study, disability was defined as a functional impairment in at least one activity of daily living () or instrumental activity of daily living (). Depressive symptoms were assessed by using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D10). We employed SPSS 27.0 to select independent risk factor variables associated with depression among disabled elderly individuals. Subsequently, a predictive model for depression in this population was constructed using R 4.3.0. The model's discrimination, calibration, and clinical net benefits were assessed using receiver operating characteristic (ROC) curves, calibration plots, and decision curves.
RESULTS: In this study, 3,107 elderly individuals aged 60 years and older with disabilities were included. Poor self-rated health, pain, absence of caregivers, cognitive impairment, and shorter sleep duration were identified as independent risk factors for depression in disabled elderly individuals. The XGBoost model demonstrated superior performance in the training set, while the logistic regression model outperformed it in the validation set, with AUCs of 0.76 and 0.73, respectively. The calibration curve and Brier score (Brier: 0.20) indicated a good model fit. Moreover, decision curve analysis confirmed the clinical utility of the model.
CONCLUSIONS: The predictive model exhibits outstanding predictive efficacy, greatly assisting healthcare professionals and family members in evaluating depression risks among disabled elderly individuals. Consequently, it enables the early identification of elderly individuals at high risk for depression.
Key numbers
1,774 of 3,107
Prevalence of Depression
Total number of disabled elderly individuals with depression out of the sample size.
0.76
AUC for XGBoost Model
Area under the curve for the XGBoost model in the training set.
0.73
AUC for Logistic Regression Model
Area under the curve for the logistic regression model in the validation set.
Full Text
We can’t show the full text here under this license.
Declarations. Ethics approval and consent to participate: This study was conducted following the principles of the Helsinki Declaration and was approved by the Biomedical Ethics Committee of Peking University. All participants signed informed consent forms before their participation, which were approved by the Ethics Review Committee of Peking University (IRB00001052-11015). Consent for publication: Not applicable. Competing interests: The authors declare no competing interests.
