A Review of Machine Learning Approaches for Epileptic Seizure Prediction

Abstract

Epilepsy is a neurological disorder that causes unusual behavior, sensations, and, in some cases, loss of awareness. It is accompanied by seizures, which are intervals of unusual patterns of brain activities. Early detection or prediction of the epileptic seizure is vital for providing effective instantaneous treatment and reducing the risk of injury. This has been an active area of research, fueled by the increasing affordability of non-invasive EEG capturing devices and the fast evolvement of the machine learning algorithms. This study provides an up-to-date review of the recent epileptic seizures approaches. Special attention is directed towards the feature extraction methods and classification algorithms. The commonly-used EEG datasets and their availability are noted. The discussed approaches range from those which rely on the traditional machine learning approaches as Naive Bayes, Support Vector Machine (SVM), and Linear Discriminant Analysis (LDA); to those that benefit from the recent deep learning approaches, such as Long Short Term Memory (LSTM) and deep Convolutional Neural Network (CNN). It also includes the hybrid approaches that combine traditional and deep learning techniques, such as combining CNN with SVM. The study concludes the discussed approaches and their limitations by comparing them in terms of reported sensitivity, prediction time and false alarm rate.