Development of interpretable machine learning models for fall prediction in older adults using the CLSA comprehensive dataset

Falls are the leading cause of injury-related hospitalization and death among older adults. To date, researchers from different disciplines have investigated the fall predictive power of different biomarkers/factors that are specific to their domains. These biomarkers/factors include performance-based measures (e.g., total time to complete a timed-up-and-go test), depression, and poor nutrition. However, none of these factors have shown to be stable in the identification of fall-prone older adults across studies. This can be due to complex interactions between different risk/protective factors at the individual level, suggesting that any single factor can touch only a small part of the entire knowledge regarding this problem. By incorporating multi-domain measures (e.g., socio-economic, genetics) in the CLSA comprehensive dataset, this project aims to develop interpretable machine learning models to predict risk of falls in older adults to inform more precise intervention strategies.