Random Neural Network-Based Epilepsy Prediction Using Statistical Features

EasyChair Preprint no. 13389

Abstract

Epilepsy is a neurological disorder characterized by recurring episodes of seizures caused by abnormal electrical activity in the brain. Predicting seizures can allow early intervention by caregivers and improve patient outcomes. This paper proposes a novel Random Neural Network (RNN)-based method for prediction of epileptic seizures using feature vector extracted from each segment of EEG data. The proposed model is trained and tested using the CHB-MIT EEG database, employing a 10-fold cross-validation technique. The proposed RNN-based model, achieved an accuracy of 95.66%, sensitivity of 93.84%, and specificity of 96.17% in predicting seizure states.

Keyphrases: Epilepsy Prediction, Random Neural Network, Remote Healthcare

@Booklet{EasyChair:13389,
  author = {Syed Yaseen Shah and Hadi Larijani and Ryan M. Gibson and Dimitrios Liarokapis},
  title = {Random Neural Network-Based Epilepsy Prediction Using Statistical Features},
  howpublished = {EasyChair Preprint no. 13389},

  year = {EasyChair, 2024}}