Sleep EEG Analysis Predicts Dementia with 99.9% Accuracy via Brain Criticality
A new study shows sleep brain waves can predict cognitive decline years before symptoms appear.
Researchers used Multifractal Detrended Fluctuation Analysis (MFDFA) on sleep EEG data from the SOF cohort. They found that cognitively healthy individuals had brain dynamics significantly closer to an optimal critical state (p ≤ 0.001), while those who later developed dementia showed a shift in DFA exponents toward 1.0, suggesting a reconfiguration of scale-free neural dynamics during sleep. The method distinguishes groups with clear spatial separation via supervised UMAP, highlighting potential for automated sleep-based screening to enable earlier preventative interventions during the prodromal window.
- MFDFA analysis of NREM sleep EEG (N2/N3 stages) shows statistically significant differences (p ≤ 0.001) between healthy and future dementia groups.
- Dementia group exhibits Hurst exponent shift toward 1.0, indicating loss of optimal brain criticality—a sign of impending cognitive decline.
- UMAP visualization achieved clear separation of groups, supporting potential for automated screening using standard sleep EEG equipment.
Why It Matters
Non-invasive sleep EEG screening could detect dementia risk years early, enabling preventative treatments before cognitive decline begins.