Longevity & AgingArticle de rechercheAccès libre

Wrist Sensors Link Light Activity and Fragmented Sleep to Memory in Older Adults

In 88 healthy older Taiwanese adults, more light activity and less night waking tracked with better memory, especially after age 72. Cross-sectional only.

mercredi 18 novembre 2026 2 vues
Publié dans JMIR Aging
Older adult in a sunlit living room walking slowly with a wrist activity tracker, a faint brain-scan overlay in the background

Résumé

In a cross-sectional study of 88 cognitively healthy adults aged 60 and older in Taiwan, researchers used 12 days of wrist actigraphy plus brain MRI to see which sleep and activity measures tracked with memory. More daily light-intensity physical activity (LPA) and less wake after sleep onset (WASO) were each modestly associated with better word-list memory scores. Both links were stronger in older participants (roughly 72 and above) and were moderated by hippocampal volume. WASO was linked to less LPA, but LPA did not mediate the WASO-memory link. Because the design is cross-sectional, with a small sample and exploratory analyses, it cannot show that improving sleep continuity or adding light activity improves memory. It is a hypothesis-generating signal, not clinical guidance.

Résumé détaillé

Memory often declines subtly with age, even without dementia, and sleep and physical activity are two of the most modifiable behaviors. Few studies have measured both with wearable devices in the same older adults, or asked whether their relationship to memory changes with age.

This was a cross-sectional analysis from the ISDHA cohort in Taiwan (ClinicalTrials.gov NCT04207502). It included 88 cognitively healthy community-dwelling adults aged 60 and older (mean age 73.5, 55.7% female). Participants wore a wrist actigraph (Actiwatch 2) for up to 14 days, with 12 days analyzed, and had 3T brain MRI and the CERAD-NB memory battery. From many actigraphy variables, the authors picked those correlating with memory after age adjustment: light-intensity activity (r=0.26, P=.02) and wake after sleep onset (r=-0.251, P=.02). Total sleep time, sleep efficiency, sedentary time, moderate and vigorous activity, and light exposure were not significantly correlated. M10 activity was marginally correlated (r=0.216, P=.046).

In multivariable models adjusted for age, education, daytime sleepiness and hippocampal volume, both WASO (β=-0.044, P=.04) and LPA (β=0.042, P=.01) predicted memory. Age moderated both associations: LPA looked beneficial beyond about 73.8 years and WASO detrimental beyond about 71.1. In participants aged 72 and over, LPA (β=0.054, P=.04) and WASO (β=-0.111, P=.01) remained significant. WASO was associated with less LPA (β=-0.325, P=.03), but the indirect path through LPA was not significant, so WASO appeared to act directly. Hippocampal volume moderated both associations, which the authors read as support for the brain reserve hypothesis.

The findings suggest that sleep continuity and everyday light movement, both measurable with consumer-style wearables, may matter more for memory in older age. The authors suggest age-tailored strategies, but this study cannot show that changing either behavior improves memory.

Caveats are substantial. The design is cross-sectional, so direction of causality is unknown; poorer memory or reduced hippocampal reserve could itself fragment sleep or reduce activity. The sample is small and relatively educated, 75% reported exercising over 150 minutes weekly, and exclusion of cognitive impairment narrows generalizability. Variables were chosen by correlation screening, with many comparisons and no stated correction, which raises false-positive risk. Age thresholds were data-derived. Effects were modest, and wrist actigraphy only approximates sleep staging and activity intensity.

Principales conclusions

  • Age-adjusted, light-intensity activity correlated with better CERAD memory (r=0.26, P=.02); wake after sleep onset correlated with worse memory (r=-0.251, P=.02).
  • In adjusted regression, LPA (β=0.042, P=.01) and WASO (β=-0.044, P=.04) independently predicted memory in 88 healthy older adults.
  • Associations strengthened at age 72 and over (LPA β=0.054; WASO β=-0.111), with age moderation thresholds near 72-74 years.
  • WASO linked to lower LPA, but LPA did not significantly mediate the WASO-memory effect; hippocampal volume moderated both associations.
  • Total sleep time, sleep efficiency, sedentary time, moderate and vigorous activity, and light exposure were not significantly associated with memory.

Méthodologie

Cross-sectional analysis of 88 cognitively healthy adults aged 60 and older from the ISDHA cohort in Taiwan, using 12 days of wrist actigraphy, 3T MRI hippocampal volume and CERAD-NB memory scores. Variables were screened by age-adjusted partial correlation, then tested with multivariable regression, PROCESS moderation (age, hippocampal volume) and moderated mediation with bootstrapped CIs. Mechanism, function and lifespan were not tested; the outcome is a memory test score at one time point.

Limites de l'étude

The cross-sectional design prevents causal or directional inference, and the sample is small, well-educated, active and cognitively healthy, limiting generalizability. Variable selection by correlation screening, apparent absence of multiple-comparison correction, data-derived age cutoffs and exploratory mediation raise overfitting and false-positive concerns, and effects were modest. Actigraphy only approximates sleep and activity intensity, and the text provided was truncated before the later results.

Ce résumé vous a plu ?

Recevez les dernières recherches sur la longévité dans votre boîte de réception chaque semaine.

Saisissez votre e-mail pour vous abonner :