Back

Data-Driven Leisure-Time Physical Activity Trajectories of Over 46 Years and Their Associations with Cognition in Nonagenarians - A Cohort Study

Iso-Markku, P.; Drouard, G.; Farrahi, V.; Varjonen, A.; Vaha-Ypya, H.; Vasankari, T.; Kaprio, J.; Vuoksimaa, E.; Aaltonen, S.

2025-09-25 epidemiology
10.1101/2025.09.24.25336548 medRxiv
Show abstract

Based on short follow-up studies, greater leisure-time physical activity (LTPA) has been associated with better cognition in old age, but more longitudinal studies are needed. Our aim was to identify long-term LTPA trajectories from midlife to late old age and examine whether these trajectories are associated with nonagenarians cognition. In total, 125 participants from the NONAGINTA - Memory and Health in Nonagenarians study were included. The participants responded to health surveys of the older Finnish Twin Cohort study including LTPA at the mean ages of 45, 52, 59 and 91 years. Cognition was assessed at the mean age of 91 years (standard deviation 1.54) via telephone interview (global cognitive function, episodic memory, and semantic fluency). We identified LTPA trajectories with K-means clustering for longitudinal data and we used generalized estimating equations models to investigate differences in cognition among the LTPA trajectories. Covariates included age, sex, and education. We found three LTPA trajectories from midlife to nonagenarian age. The largest proportion of participants belonged to the Constant low trajectory (52%), characterized by a stable low level of physical activity throughout the follow-up. Two other trajectories were Starting low and increasing (25%) and Starting high and decreasing (23%). Cognitive measures at the mean age of 91 did not differ among the LTPA trajectories. To conclude, we found three different trajectories of LTPA from midlife to late old age within a period of 46 years. There were no significant differences in cognition by distinct LTPA trajectories when age, sex, and education were considered.

Published in The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences · not in our set (fewer than 10 published preprints to learn from) · training set

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.