Skip to content

A higher sleep-based brain-age gap relative to calendar age was linked to dementia

Across community cohorts, a higher brain-age gap from overnight recordings accompanied later dementia. Its value for early detection needs evaluation.

By 100HP editorialAbstract-based explanation checked

Based on the published abstract. The full paper may contain additional methods, results and limitations.

The 30-second takeaway

A higher sleep-based brain-age gap was associated with greater dementia risk during follow-up, including after additional statistical adjustments. This suggests a possible research marker, but the observational findings cannot establish cause and effect. They also do not show whether the index can accurately identify people who will develop dementia.

Keep in mind

This abstract-based account cannot establish clinical usefulness: it does not report prediction accuracy. The authors call for evaluation of the index as a potential marker for early dementia detection.

THE NUMBERS, WITH CONTEXT

What researchers found

HR 1.39 (95% CI 1.21–1.59)

New dementia or probable dementia

Per 10-year higher index, the adjusted dementia hazard was 1.39 times as high. This compares the modeled occurrence of dementia during follow-up, not absolute probability or percentage-point change. Shared follow-up duration was unreported.

JAMA Netw Open, 2026 · Original source ↓

What the index represents

Researchers used machine learning to estimate brain age from overnight recordings of sleeping brain activity at home. The index expresses estimated brain age relative to calendar age, rather than estimated age alone.

Accounting for other factors

The models accounted for death as an event that could prevent dementia from being observed. Associations persisted after further adjustments for medical conditions, sleep-related breathing measures and a genetic factor.

CHECK THE ORIGINAL

The original publication

Machine Learning-Based Sleep Electroencephalographic Brain Age Index and Dementia Risk: An Individual Participant Data Meta-Analysis.

Sun H, Milton S, Fang Y et al.
JAMA Netw Open · 2026

PubMed ID
41854616
Record checked

Linked publication notices

AI-assisted research and writing. This explains one selected publication; it is not a complete review of everything known. Our approach.