Determinants and risk prediction models for frailty among community-living older adults in eastern China

The purpose of this study is to develop predictive models for frailty risk among community-dwelling older adults in eastern China using machine learning techniques. This approach aims to facilitate early detection of high-risk individuals and inform the design of tailored interventions, with the ultimate goals of enhancing quality of life and mitigating frailty progression in the older adult population.

This study involved 1,263 participants aged 60 years or older, who were selected through stratified cluster sampling. Frailty was assessed using the Tilburg Frailty Indicator (TFI), which encompasses physical, psychological, and social dimensions. Predictive models were constructed using decision trees, random forests, and XGBoost algorithms, implemented in R software (version 4.4.2). The performance of these models was evaluated using metrics such as the area under the receiver operating characteristic curve (AUC), ROC curves, and confusion matrices.

The results showed that 64.77% of the older adult were physically weak. Body mass index (BMI), living arrangements, frequency of visits and smoking status are the main factors contributing to frailty. When comparing predictive model metrics, random forest and extreme Gradient Lift (XGBoost) outperform decision tree models in terms of accuracy and applicability.

Older adults living in communities in eastern China showed slight frailty, and many factors influenced their frailty scores. Random forest and XGBoost models outperform decision tree models in predicting frailty in older adults, so identifying high-risk individuals early and developing personalized interventions can help slow the development of frailty and improve quality of life in older adults.