Brain Waves Measured During Sleep Predict Cognitive Impairment Years Before Symptoms Appear, Study Finds

Mass General Brigham researchers developed an AI tool to accurately predict which patients will develop cognitive impairment over the next five years by analyzing brain wave patterns during sleep.

Mass General Brigham researchers have developed an AI tool that analyzes brain wave activity recorded during sleep using electroencephalography (EEG), a non-invasive technique that measures electrical activity in the brain through sensors placed on the scalp. The AI tool was developed using sleep study data from a group of women over 65, who were tracked for five years. The researchers identified subtle differences in brain wave patterns that predicted which participants would later be diagnosed with cognitive impairment. Published in the Journal of Alzheimer's Disease, the study suggests that wearable EEG devices could help identify individuals at risk for dementia, paving the way for earlier interventions.

"Using novel sophisticated analyses, advanced information theory tools, and AI, we can detect subtle changes in brain wave patterns during sleep that signal future cognitive impairment, offering a window of opportunity for intervention years before symptoms appear," said lead author Shahab Haghayegh, PhD, a member of the Department of Anesthesia, Critical Care, and Pain Medicine at Massachusetts General Hospital, a founding member of the Mass General Brigham healthcare system, and Harvard Medical School.

Haghayegh and colleagues analyzed data that was collected for a separate trial on fracture risk in women. For that study, women aged 65 and older underwent a series of cognitive tests around the same time they participated in a sleep study, which included an overnight EEG. Researchers focused on 281 of the participants who had normal cognitive functioning at the time of the initial sleep study and repeated the same cognitive tests five years later. By that second series of assessments, 96 of these women had developed cognitive impairment.

They then applied state-of-the-art information theory techniques to extract brainwave patterns from the EEG data collected during the sleep study. Using AI, the researchers found that among people who went on to show signs of cognitive impairment, there were changes in subtle brain wave features before symptoms occurred, especially in gamma band frequencies in deep sleep. The AI tool successfully identified 85% of individuals who later developed cognitive impairment, with an overall accuracy of 77%.

The researchers say these findings could help identify patients years before cognitive impairment takes place and suggest a future where wearable EEG devices could become a powerful tool in predicting cognitive decline. Early detection could provide valuable time for individuals to access treatments and make lifestyle modifications that may help maintain cognitive health.

The authors note that further research is necessary to validate these findings across broader populations, including in males and more diverse populations, to verify the link between altered gamma wave activity during deep sleep and future cognitive impairment.

A limitation of the study is that it relied on data from a single night of sleep. However, senior author Kun Hu, PhD, physiologist in the Division of Sleep and Circadian Disorders at Brigham and Women’s Hospital and Harvard Medical School, notes that EEG data from multiple nights of sleep could be even more predictive of future cognitive impairment.

“The new, FDA-approved treatments for Alzheimer’s disease are effective at the earlier stages of dementia, but not the more advanced stages,” said Hu. “Interventions that are performed even earlier — before someone shows signs of cognitive decline — are likely to be even more effective.”

The research also opens up another exciting possibility, suggesting that manipulating brain electrical activity during sleep could reduce the risk of cognitive decline. Haghayegh and Hu are currently designing a clinical study that will assess if electrical stimulation can alter EEG patterns in sleep and thereby slow down cognitive decline.

"What makes this research particularly significant is how we can identify those at risk using a simple overnight EEG recording," says Haghayegh. "This could completely change how we approach dementia prevention."

Authorship: In addition to Haghayegh and Hu, Mass General Brigham authors include Susan Redline. Additional authors include co-first author Ruben Herzog, David A. Bennett, Kristine Yaffe, Katie L. Stone, and Agustin Ibáñez.

Disclosures: Redline reports support from the National Institutes of Health and consulting fees from ApniMed Inc, Eli Lilly Inc, and Jazz Pharma, unrelated to this project. Haghayegh reports receiving consulting fees from Achaemenid LLC, unrelated to this project. Hu is an Editorial Board Member of this journal but was not involved in the peer-review process of this article nor had access to any information regarding its peer-review.

Funding: The Study of Osteoporotic Fractures (SOF) is supported by National Institutes of Health funding. The National Institute on Aging (NIA) provides support under the following grant numbers: R01 AG005407, R01 AR35582, R01 AR35583, R01 AR35584, R01 AG005394, R01 AG027574, R01 AG027576, and R01 AG026720. KH was partially supported by the National Institutes of Health grants (RF1AG059867 and RF1AG064312). Haghayegh was supported by the National Institute of Health grants (1K99AG083234-01A1, 1R01AG083799-01, RF1AG064312, and 5T32HL007901-25). Bennett was supported by R01AG17917. Yaffe was supported in part by R35 AG071916.

Paper cited: Haghayegh S and Herzog R et al. “Predicting future risk of developing cognitive impairment using ambulatory sleep EEG: Integrating Univariate Analysis and Multivariate Information Theory Approach” Journal of Alzheimer’s Disease DOI: 10.1177/13872877251319742