Alzheimer's Disease (AD) is a rapidly growing neurodegenerative disorder that severely impairs cognitive function, particularly among older adults. Early detection is critical for timely intervention and effective management. While electroencephalography (EEG) provides a non-invasive, cost-effective tool with high temporal resolution for diagnosing Alzheimer's Disease (AD), traditional EEG-based approaches often struggle to extract informative features accurately from complex brain signals, thus limiting their diagnostic performance. This study addresses these challenges by proposing a novel artificial intelligence-based framework that integrates feature fusion with a Convolutional Long Short-Term Memory (Conv-LSTM) architecture. The model combines spectral features and deep learning-derived representations into a unified feature set, which is then processed by the LSTM network to capture both spatial and temporal patterns associated with AD. The proposed method achieves a classification accuracy of 99.8%, outperforming existing techniques and demonstrating its effectiveness in distinguishing between multiple stages of AD. Experimental results confirm the superiority of the fusion-based approach in learning high-level, discriminative features from EEG signals. This research represents a significant advancement in EEG-based Alzheimer's disease (AD) detection, with strong potential for enhancing early diagnostic systems. It supports the development of intelligent, scalable clinical tools for dementia screening and contributes to the broader effort to integrate AI into the diagnosis of neurodegenerative diseases.
Hemalatha et al. (Sun,) studied this question.