Summary: Researchers have developed a machine-learning model that can estimate a person’s “brain age” simply by analyzing their sleep.
The study reveals that when a person’s brain waves appear “older” than their chronological age, their risk of developing dementia spikes significantly. This breakthrough suggests that sleep recordings could serve as a non-invasive, early-warning system for cognitive decline years before symptoms appear.
Key Facts & Statistics
- The “Brain Age” Gap: For every 10-year increase in estimated brain age relative to actual age, the risk of dementia rose by nearly 40%.
- Protective Youthfulness: Conversely, participants whose brain waves appeared “younger” than their actual age had a significantly lower risk of dementia.
- Large-Scale Data: The study analyzed EEG recordings from approximately 7,000 participants (ages 40–94) across five different cohorts, following them for up to 17 years.
- Invisible Patterns: Conventional sleep metrics (like total sleep time or sleep stages) showed no link to dementia. The risk was only visible through 13 microstructural features of brain waves identified by AI.
- Critical EEG Markers:
- Delta Waves: Associated with deep, restorative sleep.
- Sleep Spindles: Fast bursts of activity crucial for memory consolidation.
- Kurtosis: Sudden large spikes in brain activity that, surprisingly, were associated with a lower risk of dementia.
Source: UCSF
A machine-learning analysis of brain waves recorded during sleep may help identify people at high risk of developing dementia, according to a study led by UC San Francisco and Beth Israel Deaconess Medical Center in Boston.
The study found that when a person’s “brain age,” estimated from sleep signals using EEG, exceeded their actual age, the risk of dementia increased.
For every 10-year increase in brain age versus actual age, dementia risk rose by nearly 40%. Conversely, if brain age was lower than actual age, dementia risk was lower.
The study publishes in JAMA Network Open on March 19.
Researchers used a machine-learning model that integrates 13 microstructural features of brain waves from EEG recordings. Data came from approximately 7,000 participants who had been enrolled in five studies.
The participants were aged between 40 and 94 and none had dementia at the start of the study. They were followed between 3.5 and 17 years, during which time about 1,000 participants developed the disorder.
The researchers found that analyzing fine-scale patterns in brain waves during sleep provided insights that conventional sleep metrics often miss. Earlier pooled analyses of several participant cohorts found no significant links between dementia risk and traditional sleep measures, such as time spent in different sleep stages or overall sleep efficiency.
“Broad sleep metrics don’t fully capture the complex multidimensional nature of sleep physiology,” said senior author Yue Leng, MBBS, PhD, associate professor of psychiatry at the UCSF School of Medicine.
Brain-Wave Patterns Linked to Cognitive Health
Several sleep EEG patterns that contributed to brain age are known to play roles in brain health and memory. These include delta waves, which form a rolling wave pattern associated with deep sleep, and sleep spindles – short, fast-frequency brain activity associated with memory consolidation.
Among the most notable finding was that sudden large spikes seen on EEG, known as kurtosis, were associated with a lower risk of dementia.
The researchers also found that the relationship between “older” brain age and dementia risk remained significant after accounting for factors such as education, smoking, body mass index, and physical activity, as well as other health conditions and genetic risk factors.
Potential for Early Detection
Because sleep EEG signals can be collected noninvasively, the researchers said that brain age could eventually help detect dementia risk in nonclinical settings, such as by using wearable technologies.
“Brain age is calculated from sleep brain waves,” said Leng. “We know that brain activity during sleep provides a measurable window into how well the brain is aging.”
The findings also raise the possibility that improving sleep health could influence brain aging. Leng noted that earlier studies have found treating sleep disorders can change sleep-related brain-wave patterns.
“Better body management, such as lowering body mass index and increasing exercise to reduce the likelihood of apnea, may have an impact,” said first author Haoqi Sun, PhD, assistant professor of neurology at Beth Israel Deaconess Medical Center, who developed the model with two co-authors*. “But there’s no magic pill to improve brain health.”
Co-authors:* Robert J. Thomas, MD, and M. Brandon Westover, MD, PhD, of Beth Israel Deaconess Medical Center, developed the machine-learning model with Sun. For other authors, please see the paper.
Funding: National Institutes of Health (R01NS102190, R01NS102574, R01NS107291, RF1AG064312, RF1NS120947, R01AG073410, RF1AG064312, R01NS102190, R01AG062531); National Institute on Aging (R21AG085495 and R01AG083836); National Science Foundation (2014431); National Health and Medical Research Council (GTN2009264); American Academy of Sleep Medicine.
Key Questions Answered:
A: Not necessarily. It means your brain’s electrical activity during sleep is showing signs of accelerated aging or wear. The study found this pattern could be detected in people who were cognitively healthy at the time, serving as a “predictive window” rather than a diagnosis of current disease.
A: During sleep, the brain performs critical “housekeeping,” including clearing out toxic proteins like amyloid-beta that are linked to Alzheimer’s. If brain waves are sluggish or disorganized, it may indicate that these restorative processes are failing, leading to the long-term accumulation of damage.
A: It’s not just about the quantity of sleep, but the quality of the brain waves. While there is “no magic pill,” researchers suggest that managing health factors like BMI and exercise—which reduce conditions like sleep apnea—can help preserve healthier brain-wave patterns and potentially slow down brain aging.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this sleep and dementia research news
Author: Suzanne Leigh
Source: UCSF
Contact: Suzanne Leigh – UCSF
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Machine Learning–Based Sleep Electroencephalographic Brain Age Index and Dementia Risk: An Individual Participant Data Meta-Analysis” by Haoqi Sun, Sasha Milton, Yi Fang, Hash Brown Taha, Shreya Shiju, Robert J. Thomas, Wolfgang Ganglberger, Matthew P. Pase, Timothy Hughes, Shaun Purcell, Susan Redline, Katie L. Stone, Kristine Yaffe, M. Brandon Westover, and Yue Leng. JAMA Network Open
DOI:10.1001/jamanetworkopen.2026.1521
Abstract
Machine Learning–Based Sleep Electroencephalographic Brain Age Index and Dementia Risk: An Individual Participant Data Meta-Analysis
Importance
Microstructures of sleep electroencephalography (EEG) are closely related to cognition and undergo age-dependent changes. However, their multidimensional nature makes them challenging to interpret using conventional approaches. The machine learning–based EEG brain age index (BAI) measures the deviation between sleep EEG-based brain age and chronological age.
Objective
To determine the association between sleep BAI and incident dementia in community-dwelling populations.
Data Sources
For this individual participant data (IPD) meta-analysis, sleep study data from 5 community-based longitudinal cohorts were pooled. These cohorts included the Multi-Ethnic Study of Atherosclerosis (MESA; 2010-2013), the Atherosclerosis Risk in Communities (ARIC) study (1987-1989), the Framingham Heart Study–Offspring Study (FHS-OS; 1995-1998), the Osteoporotic Fractures in Men Study (MrOS; 2003-2005), and the Study of Osteoporotic Fractures (SOF; 2002-2004).
Study Selection
Adults (aged ≥18 years) without dementia at the time of polysomnography were included.
Data Extraction and Synthesis
The BAI was computed using interpretable machine learning, incorporating sleep EEG features extracted from central channels in overnight, home-based polysomnography. Fine-Gray models were used to assess the association between BAI and incident dementia within each cohort, accounting for death as a competing risk. Cohort-specific estimates were then pooled using random-effects meta-analysis. Analyses were performed between March 2024 and September 2025.
Main Outcomes and Measures
Incident dementia or probable dementia was determined in each cohort, with death as a competing risk.
Results
This meta-analysis included 7105 participants from the MESA (n = 1802; mean [SD] age, 69.3 [9.0] years; 956 females [53.1%]), ARIC (n = 1796; 62.5 [5.7] years; 918 females [51.1%]), FHS-OS (n = 617; 59.5 [8.9] years; 318 females [51.5%]), MrOS (n = 2639 males [100%]; 76.0 [5.3] years), and SOF (n = 251 females [100%]; 82.7 [2.9] years) cohorts. The median (IQR) time to dementia was 4.8 (4.2-5.6) years in the MESA cohort (n = 119 [6.6%]), 16.9 (14.9-19.8) years in the ARIC cohort (n = 354 [19.7%]), 13.1 (8.5-16.2) years in the FHS-OS cohort (n = 59 [9.6%]), 3.6 (1.3-7.1) years in the MrOS cohort (n = 470 [17.8%]), and 4.6 (4.2-5.2) years in the SOF cohort (n = 86 [34.3%]).
Across the cohorts, each 10-year increase in BAI was associated with a 39% higher risk of incident dementia (hazard ratio [HR], 1.39 [95% CI, 1.21-1.59]; P < .001) after adjustment for covariates. These associations remained after additional adjustment for comorbidities and apnea-hypopnea index scores (HR, 1.31 [95% CI, 1.14-1.50]; P < .001) and apolipoprotein E ε4 (HR, 1.22 [95% CI, 1.02-1.45]; P = .03), and they were consistent across sex and age groups.
Conclusions and Relevance
In this IPD meta-analysis, a higher sleep EEG-based BAI was associated with a higher risk of incident dementia. These findings highlight the need to evaluate the predictive value of the BAI as a noninvasive digital marker for early detection of dementia in community settings.
