American Journal for Young Scientists · ISSN 3068-9015
Synchronization Dynamics in Dementia: A Kuramoto Model Analysis of EEG Activity in Alzheimer’s Disease and Frontotemporal Dementia
Authors: Ryan George
Published: July 20, 2026
Category: Computational Neuroscience
Abstract
Alzheimer’s disease (AD) and frontotemporal dementia (FTD) are major neurodegenera- tive disorders that produce overlapping clinical symptoms but affect partially distinct neural systems. I investigate whether resting-state electroencephalography (EEG) synchronization dynamics can help distinguish AD, FTD, and cognitively normal healthy controls (CN). Using a publicly available clinical EEG dataset of 88 subjects, EEG activity was analyzed across five canonical frequency bands: Delta, Theta, Alpha, Beta, and Gamma. Each scalp elec- trode was modeled as an oscillator in a Kuramoto synchronization framework, allowing both global phase synchronization and pairwise electrode coupling to be quantified. Spectral band power, Kuramoto order parameters, electrode coupling matrices, and machine-learning classi- fication were used to identify group-level differences. AD patients showed elevated band power across all frequency bands, consistent with diffuse cortical slowing and abnormal neural ex- citability, while FTD patients showed the lowest power across bands, suggesting more focal neural disruption. Kuramoto analysis revealed that healthy controls had stronger synchroniza- tion in slower bands, whereas both AD and FTD showed abnormal hyper-synchronization in faster Alpha and Beta bands. Electrode-coupling analysis further showed a shift from frontal- dominated connectivity in healthy controls to posterior-dominated coupling in disease groups. Machine-learning models classified AD, FTD, and CN subjects above chance, with a tuned support vector machine achieving the highest balanced accuracy of 55.6%. These findings suggest that Kuramoto-based EEG synchronization features capture meaningful differences in neurodegenerative brain dynamics and may provide useful biomarkers for distinguishing dementia subtypes. GitHub: https://github.com/ryangeorgekj/Synchronization-Dynamics-in-Dementia