Deviations from healthy brain aging, which ultimately lead to dementia, may be detected by using the electroencephalography-based brain age index (BAI) measured during sleep. These findings, from a retrospective cross-sectional study, were published in JAMA Network Open.

Polysomnograms from individuals (N=5144) acquired during 2009 and 2017 at the Sleep Laboratory at Massachusetts General Hospital were analyzed for this study. Scans (diagnostic, full-night continuous, or split-night continuous positive airway pressure) were annotated in 30-second durations as non-rapid eye movement (REM) stages 1-3 and REM. With the annotated scans, the BAI was calculated using a generalized linear model.

Participants were aged mean 54 (interquartile range [IQR], 43-65) years and 59% were men. Participants had dementia (n=81), mild cognitive impairment (n=44), were symptomatic (n=1075; defined as having some dementia-related keyword in their medical history), or were categorized as nondementia (n=2336; defined as no evidence of psychiatric disease in their medical history).

Comparing the dementia and nondementia groups, psychotic (19% vs 6%; P <.001), mood (64% vs 40%; P <.001), and anxiety (52% vs 31%; P <.001) disorders were more common, respectively.

Continue Reading

BAI was observed to increase from 0.20 (standard error [SE], 0.42) among those in the nondementia group, to 0.58 (SE, 0.41) among the symptomatic and 4.18 (SE, 1.02) among participants with dementia. Using pairwise comparisons, those with dementia had higher average BAI scores than all other groups (P <.001).

With the information extracted from the polysomnograms, BAI (odds ratio [OR], 1.44; 95% CI, 1.43-1.45; P <.001) and sleep fragmentation index (OR, 0.32; 95% CI, 1.31-1.32; P =.03) best predicted dementia.

BAI was positively correlated with dementia (regression coefficient, 4.36; SE, 2.20; P <.001), among men (regression coefficient, 2.67; SE, 0.63; P <.001), with Black ethnicity (regression coefficient, 2.17; SE, 1.50; P =.005), psychotic disorder (regression coefficient, 1.55; SE, 0.91; P <.001), cardiovascular disease (regression coefficient, 1.22; SE, 0.79; P =.002), apnea-hypopnea index (regression coefficient, 0.87; SE, 0.30; P <.001), smoking (regression coefficient, 0.73; SD, 0.64; P =.02), and periodic limb movement (regression coefficient, 0.36; SD, 0.31; P =.02).

This study was potentially limited by the fact that sleep stage scoring was manually quantified although automated methods may have been more accurate.

The study authors concluded that BAI may be an important biomarker for detecting deviations from healthy brain aging. These deviations are more likely to lead to a dementia diagnosis, indicating that BAI may be an effective tool for early detection. However, further development and testing of this method are needed before they can be applied in clinical practice.

Disclosure: Multiple authors declared affiliations with industry. Please refer to the original article for a full list of disclosures.


Ye E, Sun H, Leone MJ, et al. Association of sleep electroencephalography-based brain age index with dementia. JAMA Netw Open. 2020;3(9):e2017357. doi: 10.1001/jamanetworkopen.2020.17357