- Studies looking at the association between sleep patterns and future dementia risk have shown mixed results.
- A recent study used brain wave patterns recorded by an electroencephalogram (EEG) during a sleep study to estimate the brain age of over 7,000 participants.
- The study found that having a brain age older than actual chronological age by 10 years was associated with an 39% higher future risk of dementia.
- These findings suggest that brain age, assessed based on sleep-EEG patterns, could be potentially used to screen individuals for dementia in the future.
A recent study published in
It found that more rapid brain aging relative to actual chronological age was associated with an increased risk of dementia.
Previous studies investigating the association between broader sleep patterns, like sleep quantity and sleep quality, have yielded inconsistent results.
In contrast, the present study assessed the granular differences in sleep wave patterns that are more closely linked to brain function and dementia risk.
Study co-author Yue Leng, PhD, an associate professor of psychiatry at the University of California, San Francisco, told Medical News Today that:
“This study shows that sleep is not just restorative—it also provides a powerful window into brain health. By analyzing brain activity during sleep, we can estimate a person’s ‘brain age,’ which may reveal whether the brain is aging faster or slower than expected.”
“This study goes beyond conventional sleep measures like sleep stages or sleep efficiency, which have often shown weak or inconsistent links with dementia risk. Instead, it uses richer EEG [electroencephalogram] microstructure to generate a single, interpretable marker,” added another co-author, Matthew Pase, PhD, a professor at Monash University in Melbourne, Australia.
Christopher Allen, MD, a sleep specialist who was not involved in this research, told MNT that “this study supports the idea that sleep is not just a symptom of brain health decline but may also be a measurable early marker of neurodegenerative risk.“
“At the same time,“ Allen cautioned, “it is not ready to be interpreted as a standalone diagnostic tool. The next step is validation in more practical settings and determining how this type of sleep biomarker might complement other dementia risk markers in real-world care.”
Lack of sufficient sleep and
However, similar studies using global or macro-level sleep metrics, such as total sleep duration, sleep quality, and time spent in the different phases of the sleep cycle, have failed to show a consistent association between sleep patterns and cognitive function.
One component of a polysomnography test, commonly known as a sleep study, involves measuring brain waves using an electroencephalogram.
An electroencephalogram (EEG) involves electrodes attached to the scalp that measure the simultaneous firing of millions of brain cells. This synchronous activity of brain cells appears as waves or oscillations.
An EEG can help assess the time spent in each sleep
Non-REM sleep involves three phases (N1, N2, and N3) of varying levels of deep sleep, whereas REM involves higher brain activity levels.
While an EEG can discriminate these phases, a closer look at the EEG data reveals a significantly larger number of distinct brain wave patterns.
These microlevel sleep patterns provide a more complete picture of brain function and health. Moreover, some of these microlevel brain wave patterns have been linked to the risk of cognitive impairment.
Given the challenges associated with interpreting a large number of these brain wave patterns, the present study’s author developed a machine learning model to derive an age-like number or brain age based on multiple microlevel sleep-EEG patterns.
The machine learning model was trained on sleep data collected from a large number of individuals aged between 18 and 80 years old without psychiatric or neurological conditions.
The brain age provided by the machine learning model offers insight into how an individual’s sleep patterns differ from those of a typical healthy individual of the same chronological age.
Our bodies, along with the underlying biological processes and structures, show wear and tear with aging. While chronological age refers to a person’s actual age, biological age reflects changes in the body’s biological processes and structures.
The process of biological aging occurs at different rates among individuals, with some showing a faster or slower rate than their actual age. Biological aging is considered to be a better measure of biological function and the risk of chronic conditions than chronological age.
Consistent with this, brain age determined using imaging scans has been shown to be linked to the risk of dementia. Thus, the brain age determined from sleep EEG patterns could potentially serve as a more accessible marker of future dementia risk.
To measure the association between sleep EEG patterns and dementia risk, the authors first calculated the brain age index, which is chronological age subtracted from sleep EEG-derived brain age.
In their
In the present study, the researchers examined whether the brain ageing index could serve as an early indicator of dementia risk in a diverse community-based sample, beyond the standardized conditions of the clinic.
The study included data from 7,105 participants originally enrolled in five other independent studies examining either the long-term risk of cardiovascular disease or osteoporosis.
These studies involved a home-based polysomnography (sleep) test using a standardized protocol. Only participants who were cognitively healthy at the time of the sleep study were included in the analysis.
The cognitive status of the participants was determined at regular follow-ups based on either neuropsychological tests, physician diagnosis, and the use of medications or hospitalization for dementia.
The researchers used their validated machine learning model to estimate each participant’s brain age index from their sleep EEG data.
A 10-year increase in the brain age index was associated with a 39% higher risk of dementia during the follow-up period. This association remained intact after accounting for variables such as education, physical activity levels, age, and sex.
In addition, the brain age index was linked to future dementia risk after adjusting the statistical analysis for factors, such as baseline cognitive status, genetic predisposition for Alzheimer’s disease due to the APOE e4 gene variant, and comorbidities, such as stroke, cardiovascular conditions, and sleep apnea severity, that were assessed at the time of the sleep study.
David Jones, MD, a neurologist at the Mayo Clinic, who was not involved in the current research, spoke to MNT about the findings.
Jones noted that a previous study analyzing conventional sleep metrics, such as sleep duration and time in different sleep stages, using data from the same five studies, failed to find an association between sleep metrics and dementia risk.
“The fact that the microstructural approach succeeds where the macrostructural one fails tells us something important: The brain’s vulnerability to neurodegeneration is written in the detailed texture of sleep, not just its broad outline.”
The strengths of the study included the large and diverse sample of more than 7,000 participants from a community-setting. In addition, the study obtained an easily interpretable measure of brain aging from complex sleep EEG data using machine learning.
“A key strength of the study lies in its strategic application of machine learning algorithms to large-scale population data, enabling clinically meaningful interpretation of sleep EEG outputs,“ Arman Fesharaki-Zadeh, MD, PhD, an assistant professor of psychiatry and of neurology at Yale University, explained.
“This noninvasive methodology holds substantial promise for real-world clinical implementation. Given the high prevalence of sleep disorders and the large volume of sleep studies conducted annually in the U.S., this approach is inherently scalable,” suggested Fesharaki-Zadeh, who was not involved in the current research.
“Polysomnography is already used widely for sleep disorders, and home-based recordings are becoming more common,“ Jones also said.
“If this measure can be validated in wearable EEG devices, it could become a genuinely scalable, noninvasive tool for early dementia risk stratification, complementing emerging blood-based and imaging biomarkers,“ he aded, “although a better mechanistic and biological understanding would accelerate these efforts.”
However, the study had a few limitations. For instance, the methods used to diagnose dementia among participants and the durations of follow-up visits were not consistent across the five original studies.
Jones cautioned that “the study pools all-cause dementia, so we don’t know whether this marker is equally predictive across Alzheimer’s disease, Lewy body dementia, vascular dementia, and other subtypes, a distinction that matters considerably for clinical decision-making.”
He also highlighted that the study was observational in nature and does not establish that sleep patterns associated with increased brain age cause dementia.
