Title: Prediction of Functional Disability in Older Chinese Adults Using a Random Survival Forest Model

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background As life expectancy increases, so does the risk of age-related diseases and functional disability, which significantly raises the risk of all-cause mortality in older adults. Individuals with disabilities may die up to 20 years earlier than those who are non-disabled. Objectives To develop a prediction model for functional disability using random survival forest analysis (RSF). Methods Data were drawn from 2,364 older adults without functional disability from the China Health and Retirement Longitudinal Study (CHARLS), conducted from 2011 to 2020. Functional disability was the primary outcome. Univariable and multivariable Cox regression analyses were used to identify significant factors, which were then screened using variable importance (VIMP) and minimal depth to construct the RSF model. The model's performance was evaluated using calibration curves and the area under the receiver operating characteristic (AUC) curve. Multimorbidity trajectories were also identified as potential risk factors through group-based multi-trajectory modeling. Results Four multimorbidity trajectories were identified: no multimorbidity, newly-developing, moderate-developing, and severe-developing. The RSF model outperformed the Cox regression model in predicting functional disability, with key factors including age, education, walking time, grip strength, CES-D score, and multimorbidity trajectories. Significant factors identified were CES-D score, grip strength, multimorbidity trajectory, age, and the use of antihypertensive medications. Conclusions The RSF model, based on CHARLS data, effectively predicts functional disability in older adults, with depressive symptoms, handgrip strength, multimorbidity trajectories, age, and antihypertensive medication use emerging as key predictors.

Article activity feed