Written by Erin Yeh
Sleep disturbances appear as early indicators of dementia. However, the broader construct of sleep shows inconsistent associations between cognitive impairment and incident dementia. These broad sleep measures do not fully reflect the complex and multidimensional nature of sleep physiology. according to Yu Ling, MBBS, PhD, is an associate professor of psychiatry at the University of California, San Francisco School of Medicine. Fortunately, sleep electroencephalography (EEG) can capture detailed patterns of brain activity that directly reflect basic neurological processes and functions, detecting dementia before symptoms appear while providing insight into how cognitive abilities change with aging.
Changes in EEG microstructure usually follow a predictable pattern as people age, but some individuals may show a more rapid decline in these patterns. Although previous studies have shown an association between cognitive impairment and multiple EEG patterns during sleep (such as spectral power, sleep depth, and spindle slow oscillation coupling), the vast amounts of EEG patterns are difficult to summarize and interpret.
To overcome this barrier, Ling and her team developed a sleep EEG-based brain age using a first-of-its-kind interpretable machine learning approach that integrates multiple age-dependent EEG microstructures into a single age-like number. They then calculated the brain age index (BAI) – the difference between brain age and chronological age – using electroencephalogram (EEG) microstructures during sleep. They investigated the association between the BAI and dementia risk across five groups and whether key dementia risk factors influenced this association. Their findings were published in The JAMA Network is open (DOI: 10.1001/jamanetworkopen.2026.1521).
The relationship between BAI and incident dementia
The study included five cohorts from the Multi-Ethnic Study of Atherosclerosis with 956 females and 846 males; The Atherosclerosis Risk in Communities Study of 918 females and 878 males; Framingham Heart Study – Offspring study with 318 females and 299 males; Study of osteoporotic fractures in men with 2639 males; A study of osteoporotic fractures with 251 females. The team noted that the majority of participants in all groups were cognitively normal at the time of the sleep assessment.
While a large participant pool means a diverse population, it also means a combination of different demographic characteristics, data collection methods, dementia assessment, and follow-up periods, which may introduce heterogeneity and cause potential bias when combining results from all studies.
The results found that 1,082 participants developed dementia. Across cohorts, each 10-year increase in BAI was associated with a 39% increased risk of incident dementia. Even after taking into account factors such as age, sex, race and ethnicity, use of sleep medications, physical activity level, education, smoking status and body mass index, the risk remained the same at 39%. The team also adjusted for cognitive scores, diabetes, high blood pressure, heart attack, stroke, depression, and the Apnea-Hypopnea Index score at the time of the sleep assessment, and the association barely decreased. These findings may indicate that the relationship between BAI and subsequent dementia is separate from cognitive status and several comorbidities.
Brain age and dementia risk
Previous studies conducted by the team have confirmed the existence of sleep-based BAI in clinical settings. The results of the current study show, for the first time, that the association of BAI with future dementia risk applies to people outside clinical populations. Instead of predicting dementia, BAI is trained on large lifetime EEG datasets using a person’s actual age as a reference. This allows the use of a much larger and more diverse data set than would be available for dementia-focused models. As such, the findings reinforce the potential of the BAI as a useful marker of accelerated brain aging.
Other biological mechanisms may explain why high BAI is associated with increased risk of dementia, such as high levels of tau proteins in the cerebrospinal fluid and increased amyloid buildup associated with changes in sleep-related brain activity. Ling and her team found this convenient APOE Genotype had little to no relationship with the relationship between BAI and dementia risk, suggesting that BAI does not solely reflect genetic risk for Alzheimer’s disease.
One limitation of the study is that only death was included as a competing risk for dementia. Other life events, such as major surgeries, psychiatric illnesses, or acute medical conditions, may influence follow-up or ascertainment of dementia and warrant consideration as additional competing risks.
Additionally, because the study is observational, it cannot infer a causal relationship between BAI and dementia. As a composite measure, the BAI is not a reasonable treatment target and should be considered a prognostic marker for future dementia risk. If an individual shows an older BAI than his or her age would indicate, it is recommended to examine specific EEG microstructural components to determine what drives this deviation, interpret their neurophysiological significance, and evaluate whether these underlying processes may represent viable therapeutic targets.
Reliance on EEG microstructural features to estimate BAI may also limit broader applicability. The team writes that future research should use wearable devices for validation to ensure broader generalizability and support clinical implementation. Not only has the BAI provided insights into neurophysiological signals that reflect future dementia risk or resilience, it may also help identify people who need closer cognitive monitoring, identify high-risk patients, and better inform clinical decision-making.
