RNNSczNet: Recurrent Neural Network with attention and ReUL for the Detection of Schizophrenia using EEG Signals

The activation function of the Rectified Linear Unit (ReLU), integrated with Recurrent Neural Networks (RNNs), GooLuNet, and other networks, has greatly improved the early identification of schizophrenia (SZ) using EEG data. RNNs' ability to capture temporal dependencies, combined with GooLuNet's inception modules for spatial feature extraction, enables accurate classification of SZ patients. ReLU is pivotal in ensuring efficient gradient propagation, reducing training time, and preventing vanishing gradient issues, thus improving model performance. The proposed model achieves a remarkable accuracy of 98.6%, outperforming more traditional machine learning techniques like Random Forest (RF) and Support Vector Machines (SVM). The architecture can also detect SZ around six months before symptoms appear, which shows that it might be used in real-time clinical settings. This combination of deep learning techniques offers a robust framework for early diagnosis, improving patient outcomes through timely intervention and treatment.