Investigation of mental health issues and brain injury using eeg rhythms

Abstract

v newlineABSTRACT newlineIntegrating Machine Learning algorithms with electroencephalography (EEG) to diagnose brain injury and mental disorders has emerged as a significant area of research in recent times. However, before machine learning algorithms can be implemented, EEG data must be appropriately preprocessed and prepared using Feature Engineering. The selection of Feature Engineering approaches can indeed distinguish between a complex and ineffective machine learning model and one that is useable. As stated otherwise, feature engineering is essential, mainly when dealing with intricate, non-stationary data like EEG. The anticipated research aims to diagnose mild traumatic brain injury and schizophrenia. In the proposed work, Multi-level Discrete Wavelet Transform (MDWT) is applied as a feature engineering technique for separating the various EEG rhythms, namely, gamma, beta, alpha, and theta. Then, statistical features are calculated from these separated EEG rhythms. The separate and/or ensembled machine learning classifiers are applied for the final disorder classification. The overall methodology of the research is as follows: (a) preprocessing, (b) EEG rhythm separation through MDWT, (c) calculation of statistical features, (d) classification of features using separate and/or ensembled machine learning classifiers. newline