Schizophrenia Detection via EEG using RNNSczNet with Attention Mechanisms and ReUL Activation

Abstract

Schizophrenia (SZ) early identification utilizing EEG data has been improved with the integration of the Rectified Linear Unit (ReLU) activation function with Recurrent Neural Networks (RNNs), GooLuNet, and other neural architectures. RNNs excel in capturing temporal dependencies, while GooLuNet’s inception modules effectively extract spatial features, enabling precise classification of SZ patients. By promoting effective gradient propagation, speeding up training, and resolving vanishing gradient issues, the ReLU activation function significantly improves model performance. The suggested framework outperforms more established machine learning techniques like Random Forest (RF) and Support Vector Machines (SVM) with an astounding accuracy of 98.6%. The architecture demonstrates the ability to identify SZ up to six months prior to symptom manifestation, highlighting its potential for real-time clinical application. This amalgamation of advanced deep learning methodologies provides a robust and innovative framework for early diagnosis, fostering improved patient outcomes through timely intervention and treatment.

Topics & Concepts