A Preliminary Study on Band-Specific Multivariate Multiscale Fuzzy Entropy of EEG in Mild Cognitive Impairment

Mild Cognitive Impairment (MCI) is a transitional phase between normal aging and Alzheimer's Disease (AD), necessitating early and precise detection for effective intervention. Recently, electroencephalography (EEG) has gained importance as a non-invasive tool to capture the complex nonlinear and multivariate nature of neural signals. In this context, this study investigates the use of Multivariate Multiscale Fuzzy Entropy (MMFE), an advanced entropy-based metric, as a biomarker for distinguishing between healthy subjects, MCI, and AD patients. Specifically, resting-state closed-eye EEG recordings were used. To enhance reliability, an amplitude normalization preprocessing step was implemented to mitigate bias due to channel variance differences. Analyses across five frequency bands, such as full-band, delta (δ), theta (θ), alpha (α), and beta (β), revealed distinct patterns of entropy alterations associated with disease progression. Specifically, significant reductions in MMFE were found in the β band within the right frontal region, where MCI patients showed lower complexity than healthy controls across multiple temporal scales, indicating early frontal dysfunction related to executive and attentional processes. Moreover, marked MMFE decreases in the δ and α bands distinguished AD from MCI groups, reflecting advanced neurodegeneration. These findings support MMFE's potential as a sensitive and cost-effective EEG biomarker for detecting and monitoring cognitive decline.