Sleep AI Determines Whether the Brain is Aging Faster Than the Person

Researchers from UC San Francisco and Beth Israel Deaconess Medical Center trained a machine learning algorithm on 13 microscopic EEG features. The model was applied to data from approximately 7,000 adults aged 40 to 94 from five independent studies. None had dementia at baseline. Over a follow-up period ranging from 3.5 to 17 years, about 1,000 people developed it.

The algorithm estimated "brain age." If it turned out to be higher than the chronological age, the risk of dementia grew noticeably. Every additional 10 years of difference increased the probability by about 40%. For those whose brains looked younger than their chronological age, the risk was lower. The association remained even after accounting for education, smoking, body mass index, physical activity, other diseases, and genetic factors.

Among the important features were deep sleep delta waves, sleep spindles (short bursts of activity associated with memory consolidation), and the kurtosis metric—sharp signal spikes that were found to be associated with a lower risk.

Since EEGs can be recorded relatively easily, in the future, such assessments could theoretically be obtained using wearable devices. For now, it is a research tool, but it shows that the sleeping brain can provide early signals about its condition years before clinical symptoms appear.

Sources:

1. University of California - San Francisco. "AI can tell if your brain is aging faster than you are." ScienceDaily (July 31, 2026).
2. Sun H. et al. "Machine Learning–Based Sleep Electroencephalographic Brain Age Index and Dementia Risk." JAMA Network Open (2026). DOI: 10.1001/jamanetworkopen.2026.1521.

Author: Maksim Aleksandrovich Erdyakov.