EEG-Based Alzheimer’s Diagnosis Using Hybrid Convolutional and Recurrent Neural Networks

Abstract

Alzheimer's disease (AD) is a neurological disorder that gets worse over time and has a big effect on brain function. It is important to get a correct diagnosis as soon as possible so that treatment can be effective. EEG, which records brain activity without touching the brain, is non-invasive, cheap, and can show activity in real time. It has become a hopeful way to find AD. By exploiting their spatial and temporal characteristics, this paper proposes a hybrid deep learning technique combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to better analyse EEG data. The CNN component removes spatial-frequency characteristics from time-frequency representations of EEG data; the LSTM component determines temporal dependence of EEG sequences on one another. The combined CNN-RNN architecture outperforms both conventional machine learning models and single deep learning systems in terms of accuracy, F1-score, and stability. Using standard EEG datasets for experiments shows that the proposed model can accurately classify things while still being easy to program. This means it can be used in clinical settings. The current state of EEG-based Alzheimer's detection is improved by this method, which also lays the groundwork for smart, real-time diagnostic tools.