Related Experiment Videos
Understanding Multilevel Correlates of Long-Term Physical Activity Trajectories Among Middle-Aged and Older Adults: A
Wenjing Liu1, Jiao Liu2, Yunru Shao1
1School of Psychology, Beijing Sport University, Beijing 100084, China.
Behavioral Sciences (Basel, Switzerland)
|July 28, 2026
Summary
Physical activity patterns in older adults vary significantly over time, not following a single path. Understanding these diverse trajectories using multilevel factors is key for tailored interventions.
Area of Science:
- Behavioral Science
- Gerontology
- Public Health
Background:
- Physical activity in later life is complex, influenced by multiple factors.
- Understanding long-term physical activity patterns is crucial for health and well-being in aging populations.
Purpose of the Study:
- To examine heterogeneous long-term physical activity (PA) patterns among middle-aged and older adults.
- To explore how multilevel variables, organized by the Social Ecological Model (SEM), contribute to classifying PA trajectories.
Main Methods:
- Utilized longitudinal data from the China Health and Retirement Longitudinal Study (CHARLS).
- Employed latent class growth modeling and growth mixture modeling to identify distinct PA trajectory groups.
- Applied machine-learning classification to evaluate the contribution of SEM-organized baseline variables to trajectory membership.
Main Results:
- Identified three distinct PA trajectories: persistently low, high-decreasing, and moderate-increasing.
- Random forest classification achieved high performance (AUC 0.943-0.945).
- Key predictors for classification included social participation, age, education, and gender.
Conclusions:
- Long-term PA patterns exhibit significant heterogeneity among older adults.
- Multilevel factors within the SEM framework are valuable for classifying distinct PA trajectories.
- Future research should consider dynamic PA trajectories and their classification-relevant features, moving beyond static activity levels.
