Alzheimer's disease with progression analysis using a novel dilated convolutional attention based long short term memory model

Alzheimer's disease (AD) is an irreversible neurodegenerative syndrome that affects memory, cognitive abilities and behaviour. Detecting AD in the early stage is crucial to improve the quality of life. However, traditional diagnostic approaches and manual analysis of neuroimaging data are slow, subjective and lead to human mistakes. Existing machine learning techniques often have difficulty in identifying complex patterns in high-dimensional biomedical data. These drawbacks emphasize the necessity for a more efficient and automated diagnostic system. This study introduced a new deep learning based hybrid framework for classifying and predict progression of AD. The method comprises three main steps: data acquisition, feature extraction and classification. Initially, EEG signals are collected from the CAU-EEG dataset. Then, features such as time domain features, frequency domain features and time frequency domain features are extracted. Finally, classification is performed by dilated convolutions attention based long short term memory (DC-ALSTM). Investigational results show that the proposed model outperforms existing baseline methods. DC-ALSTM achieved a classification 99.26% accuracy, 99.21% precision, recall at 99.23% and 99.22% F1-score, which indicates outstanding diagnostic capability.