Identifying demographic and home features influencing older adults' home functioning by using machine learning models

Objectives: This study aimed to identify demographic and housing features associated with functional difficulties experienced by older adults in their homes.

Participants: Individuals aged ≥ 65 years who completed American Housing Survey (AHS) questionnaires. We selected one random person per household and excluded participants with missing data for the 12 AHS functional challenge items.

Methods: Multiple machine learning models were compared to identify the best-performing model, which was then used to analyze the impact of demographic and housing features on older adults' functional difficulties at home.

Results: The random forest model was selected for its preferred predictive performance (accuracy: 85.8%, sensitivity: 94.4%, specificity: 60.2%, precision: 87.6%, and negative predictive value: 78.2%). The top five variables that significantly influenced the model were: 1) walking disability, 2) presence or use of a cane or walker, 3) presence or use of handrails or grab bars in the bathroom, 4) go-outside-home disability, and 5) self-care disability. These variables had a stronger impact on the model than the householder's health and age.

Conclusion: Home modifications and environmental adaptations may be critical in enhancing functional abilities and independence among older adults. These findings could inform the development of interventions that promote safe and accessible living environments for older adults, thereby improving their quality of life.