Automatic Schizophrenia Detection Using Discrete Wavelet Transform from EEG Signal

This study investigates the integration of Discrete Wavelet Transform (DWT) with both machine learning (ML) and deep learning (DL) techniques to enhance the diagnosis of schizophrenia (SZ) using EEG data. Current diagnosis relies on subjective clinical assessments, presenting challenges in differentiation from other mental disorder. DWT, with its ability to capture both temporal and spectral information, serves as a valuable feature extraction method. ML models, including Support Vector Machines (SVM), Random Forest (RF), and Gradient Boosting, demonstrate improved performance with feature reduction, achieving accuracy rates up to 77.81%. DL models, particularly Multilayer perceptron and Convolutional Neural Network (CNN), outperform ML models, with Multilayer perceptron reaching an impressive accuracy of 88.96% and CNN at 81.77%. These results highlight the potential of DL techniques for precise SZ diagnosis, paving the way for more effective treatment strategies and outcomes in the future.