A baseline analysis of a randomized controlled trial has found that wrist actigraphy, the wearable most commonly used to track sleep in aging research, showed no significant association with any of five plasma Alzheimer's disease biomarkers in older adults with insomnia symptoms, while a smaller subsample who wore ambulatory EEG sensors at home did show a link. The findings were published in The Journals of Gerontology: Series A (Volume 81, Issue 6, June 2026), by Sofia Liu, Jing Huang, Russell Calderon and colleagues, using data from a registered trial (NCT03959202).
The study drew on 103 sedentary, community-dwelling adults without dementia, averaging 70 years old and 80.6 percent female, all reporting insomnia symptoms. Participants completed a standard sleep questionnaire (the Pittsburgh Sleep Quality Index) and wore actigraphy watches; blood was drawn for plasma markers of amyloid-beta (Abeta42, Abeta40, the Abeta42/40 ratio), total tau, and neurofilament light chain, a marker of neurodegeneration. A subgroup of 56 participants also underwent two nights of ambulatory EEG, a home-based device that, unlike a wrist monitor, can distinguish REM from non-REM sleep stages. In adjusted models, none of the actigraphy measures, sleep duration, sleep efficiency, sleep onset latency, or wake after sleep onset, were significantly associated with any of the five plasma biomarkers.
By contrast, in the EEG subsample, a greater percentage of REM sleep was associated with a lower plasma Abeta40 level and a higher Abeta42/40 ratio, a pattern the researchers describe as a more favorable amyloid profile. Higher scores on the sleep questionnaire, indicating worse subjective sleep quality, were nominally linked to higher Abeta42 and higher neurofilament light, though the authors call these associations preliminary and say they need confirmation in larger studies. The authors frame the actigraphy null result and the EEG association together as evidence that stage-specific, physiologic detail, which a wrist accelerometer cannot capture, may be needed to detect these biomarker relationships in this population.
The study is cross-sectional at baseline, meaning it captures a single snapshot rather than change over time, and the EEG comparison rests on 56 people, a small subsample even within an already modest trial of 103. The authors themselves note the subjective sleep quality findings are nominal, and independent replication in larger, more diverse cohorts has not yet been done.