Interpretable Machine Learning of Nonlinear EEG Dynamics Enables Robust Differentiation of Alzheimer's Disease and Frontotemporal Dementia

This repository contains the complete computational pipeline for the study titled “Interpretable EEG-based classification of Alzheimer’s disease and frontotemporal dementia using nonlinear dynamics and explainable machine learning.” The project develops and validates a transparent machine learning framework to distinguish Alzheimer’s disease (AD), behavioral variant frontotemporal dementia (FTD), and healthy controls (HC) using resting-state electroencephalography (EEG). The methodology emphasizes nonlinear dynamic features (e.g., recurrence quantification, Lyapunov exponents, multiscale entropy) and model interpretability via SHAP and topographic mapping.