Identification of older adults’ mobility decline trajectories in longitudinal studies using machine learning approaches

Mobility limitation is a strong predictor of disability, hospitalization and death among older adults. To address new personalized rehabilitative interventions for improving late-life mobility, researchers’ attention has increasingly focused on understanding trajectories of mobility decline, rather than merely studying the onset of limitations. However, identification of these trajectories is challenging, as there are complex interactions between different factors (e.g., biological, behavioral, environmental), at the individual level, which impact the trajectories. This study will work to develop machine learning models to identify mobility decline trajectories in large-scale longitudinal datasets. By investigating a comprehensive set of measures, the anticipated models will provide information on risk factors contributing to mobility decline trajectories. Therefore, early preventive interventions could be addressed to delay the presence of adverse mobility-related outcomes (e.g., falls).