Exploring outdoor activity limitation (OAL) factors among older adults using interpretable machine learning

Area of Science:

• Gerontology
• Artificial Intelligence in Healthcare
• Public Health

Background:

• Outdoor activity limitation (OAL) in older adults is a complex issue influenced by multiple aging-related factors.
• Understanding these multidimensional constraints is crucial for maintaining functional independence in aging populations.

Purpose of the Study:

• To develop and evaluate interpretable machine learning (ML) models for predicting OAL in older adults.
• To identify the most significant multidimensional aging factors and constraints contributing to OAL.

Main Methods:

• Utilized data from 6794 community-dwelling older adults (≥65 years) from the National Health and Aging Trends Study (NHATS).
• Assessed predictors across six dimensions: sociodemographics, health, physical capacity, neurological function, daily living, and environment.
• Applied multidimensional interpretable machine learning techniques for model construction and analysis.

Main Results:

• The comprehensive multidimensional ML model achieved the highest predictive performance (AUC: 0.918).
• Physical capacity emerged as the strongest predictor (AUC: 0.895), followed by daily habits and abilities, and physical health.
• Key predictors included SPPB score, lifting ability, leg strength, self-rated health, and fear of falling.

Conclusions:

• Integrating potentially reversible factors, such as neurological performance and physical function, enhances OAL risk assessment.
• Prioritizing interventions targeting high-contribution, modifiable factors can effectively address OAL in older adults.
• ML models offer valuable insights for developing targeted, sequential interventions to mitigate OAL.