Seizure Prediction and Classification Using Affective Technology

Seizure Prediction and Classification Using Affective Technology

Folakemi Favour Kayode (Federal University of Technology, Akure, Nigeria)
DOI: 10.4018/IJICTRAME.2021010101
OnDemand PDF Download:
No Current Special Offers


Epilepsy is a neurological issue that is described by abrupt and irregular seizures. As indicated by the World Health Organisation (WHO), roughly 1% of the total populace are epileptic patients. The abrupt nature of epileptic seizures constitutes a major disabling aspect of the ailment due to the impediments in patients' daily activities. Therefore, a method that can speculate the event of seizures could essentially improve the wellbeing of epileptic patients. Hence, the point of this paper is to build up a model that can anticipate seizures and furthermore characterize them, using a GSR sensor, temperature sensor, and also a pulse rate sensor.
Article Preview

1. Introduction

Epilepsy is a neurological ailment that occurs as a result of an abrupt disruption of the brain activity which is associated with repetitive and unprovoked seizures. The abrupt nature of epileptic seizures constitutes a major disabling aspect of the ailment, due to the impediments, in patients’ daily activities thereby resulting poor standard of living. Throughout the years epilepsy has been treated through prescriptions and medical procedure, however these have a few aftermaths or symptoms as well as neglect to sufficiently control seizures of some affected patients (Shorvon, 2005). A dependable framework that can foresee seizures may amazingly guarantee the wellbeing of patients, upgrade their way of life, and furthermore increase their possibility of controlling seizures by regulating medicine as right on time as would be prudent.

Expectation of seizure depends on the speculation that a progress state (preictal) exists between the interictal (typical) and the ictal (seizure) states. There are quantities of clinical confirmations that help this speculation (Maiwald et al., 2004). As needs be, in the course of the most recent two decades, scientists have put forth extraordinary attempt in endeavouring to foresee epileptic seizures dependent on EEG signals. Conradsen (2013) proposed a framework for the Identification of Epileptic Seizures with Multi-modular Sign Handling, he introduced a strategy where the help vector machine classifier is applied on highlights dependent on wavelet groups. This was utilized on multi-modular information from control subjects. This calculation was then applied to structure a wearable gadget dependent on uni- or multi-modalities that could identify at whatever point a seizure starts and sounds an alert when the seizure is identified, which brought about the identification of just one sort of seizure. Le et al. (2001) proposed a strategy to anticipate seizures utilizing the closeness between an interictal reference and the present windows of EEG dependent on a nonlinear investigation of zero-intersection interims. They applied this calculation to the EEG signals from 23 patients with fleeting projection epilepsy (TLE) which brought about 96% affectability and a normal expectation time of 7 min. In a work detailed by Alexandros., et al., (2015) Robotized Epileptic Seizure Location Techniques, Two sorts of computerized strategies for examination of epileptic EEG accounts were accounted for in the writing: those focused on between ictal spike identification, and those focused on epileptic seizure investigation and portrayal of strange EEG exercises in long haul chronicles, anyway this work can just help the scientists in the field of EEG signal examination to comprehend and embrace the different techniques for the recognition of neurological issue related with EEG accounts. Milosevi (2015) explored the capability of programmed epileptic seizure identification in paediatric patients. He utilized component choice strategies to recognize the most applicable highlights for the differentiation between every epileptic seizure class and all other nocturnal movements utilizing ACM signals. He presumed that it is best when patient is not involved in any activity. Ihsan et al., (2018) proposed a Robotized Framework for Epilepsy Identification utilizing EEG Brain Signs dependent on Profound Learning Approach. The manual investigation of EEG brain signals which is a tedious and arduous procedure was the motivation for this work. The proposed framework depended on profound realizing, which is the best in AI approach. For this framework, a memory productive and basic pyramidal one-dimensional profound convolutional neural system (P-1D-CNN) model was presented, which is a start to finish model, and includes less number of learnable parameters. The framework will be structured as a troupe of P-1D-CNN models, which takes an EEG signal as information, passes it to various P-1D-CNN models lastly melds their choices utilizing greater part vote. To conquer the issue of little dataset, two information enlargement plans were presented for learning P-1D-CNN model. Because of less parameters, P-1D-CNN model is anything but difficult to prepare just as simple to convey on chips where memory is constrained. The restriction of this framework is that the epilepsy identification techniques identify seizures after their event.

After the review of the seizure prediction methods discussed above, there is a need to find new mechanism for the prediction of seizures. Hence the point of this paper is to build up a model that can anticipate seizures and furthermore characterize them.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 11: 2 Issues (2022): Forthcoming, Available for Pre-Order
Volume 10: 2 Issues (2021)
Volume 9: 2 Issues (2020)
Volume 8: 2 Issues (2019)
Volume 7: 2 Issues (2018)
Volume 6: 2 Issues (2017)
Volume 5: 2 Issues (2016)
View Complete Journal Contents Listing