American Journal for Young Scientists · ISSN 3068-9015
Distinct EEG Spectral Signatures Differentiate Alzheimer’s Disease, Frontotemporal Dementia, and Controls
Authors: Aarian Dasgupta, Samyak Jain
Published: October 19, 2025
Category: Computational Neuroscience
Abstract
Electroencephalography (EEG) provides a non-invasive window into the neural oscilla- tory patterns associated with neurodegenerative disorders. In this study, we compared EEG power spectral features across Alzheimer’s disease (AD), frontotemporal dementia (FTD), and healthy control participants to identify frequency band alterations and assess their utility for automated classification. Power spectral density was computed for five canonical frequency bands: delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz), and gamma (30–50 Hz). Group comparisons revealed distinct spectral signatures, including reduced al- pha power and elevated theta power in AD, altered gamma power in FTD, and higher alpha power in controls. We further evaluated the discriminative capability of these features using four machine learning algorithms—logistic regression, random forest, multilayer perceptron (MLP), and XGBoost. Classification performance varied across models, with MLP achieving the highest overall accuracy (66.14%), while random forest excelled in distinguishing controls (73.87%) and MLP performed best in classifying FTD (61.70%). These findings highlight characteristic EEG spectral alterations in AD and FTD and demonstrate the feasibility of us- ing band power features for automated differentiation between dementia subtypes and healthy aging. Our results underscore the potential of EEG-based machine learning approaches as ac- cessible, cost-effective tools for early diagnosis and disease monitoring in clinical and research settings.