Image Courtesy: Shutterstock

A new artificial intelligence system could help identify people at higher risk of developing dementia by estimating how old their brains appear based on electrical activity recorded while they sleep.

Researchers from the University of California, San Francisco, and Beth Israel Deaconess Medical Center developed a machine-learning model that analyzes brain waves captured through electroencephalography, or EEG. The system calculates a person’s “brain age” and compares it with their actual age. The larger the gap, the greater the observed dementia risk.

The study found that for every 10-year increase in the difference between estimated brain age and chronological age, the likelihood of developing dementia increased by nearly 40%, while people whose brains appeared younger had a lower risk.

The researchers built the model using 13 microscopic characteristics found in sleep-related EEG recordings. They tested it against data from roughly 7,000 people across five studies. Participants were between 40 and 94 years old and did not have dementia when their respective studies began. Follow-up periods ranged from 3.5 to 17 years, during which about 1,000 participants developed dementia.

The findings suggest that detailed patterns in sleeping brain activity may reveal changes that conventional sleep measurements miss. Previous research has found little connection between dementia risk and broad measures such as total sleep time, sleep-stage duration, and sleep efficiency.

Several of the signals used by the AI are already associated with cognitive function. Delta waves, which are prominent during deep sleep, and sleep spindles, brief bursts of rapid activity, are both believed to play roles in memory processing.

The researchers also identified an association between dementia risk and kurtosis, a statistical measure describing unusually large spikes in EEG signals. Higher levels of this feature were linked to lower dementia risk.

The relationship between older estimated brain age and dementia remained significant even after accounting for factors including education, smoking, body mass index, physical activity, other health conditions, and genetic risk.

The researchers say the approach could eventually support earlier dementia risk assessment because EEG is noninvasive and could potentially be incorporated into wearable sleep-monitoring technologies. However, the findings do not establish that an older brain age causes dementia or that changing sleep patterns will necessarily prevent it.