Title: An efficient hybrid approach for the prediction of epilepsy using CNN with LSTM

Authors: Arpita Srivastava; Anuj Singh; Arvind Kumar Tiwari

Addresses: KNIT Sultanpur, Uttar Pradesh, India ' KNIT Sultanpur, Uttar Pradesh, India ' KNIT Sultanpur, Uttar Pradesh, India

Abstract: Epileptic seizures are a severe neurological disorder with significant implications for public health. Epileptic seizure is one of the neurological disorders which affect either children in the age group of 10-20 years old or adults in the age group of 65-70 years old. It affects brain cells. Electroencephalogram (EEG) is the best tool for the recording of brain electrical activity. Epileptic seizures can be studied in four stages known as preictal, ictal, postictal, and interictal. This paper presents a literature review for the prediction of epilepsy using various machine learning-based approaches. This paper also presents the comparative analysis of various computational-based techniques used to predict epilepsy. This paper proposes a hybrid approach for the prediction of epilepsy using convolutional neural network and long-short-term memory. Here, the proposed model achieved an accuracy of 98%, precision of 98.21%, recall of 92.02%, F1-score of 95.01%, specificity of 99.56%, MCC of 93.84%, TPR of 92.02%, FPR of 0.44% and AUC is 100%. IT is also observed that the proposed model performed better in comparison to other approaches.

Keywords: epileptic seizure; convolutional neural network; CNN; long-short-term memory; LSTM; deep learning; support vector machine; SVM.

DOI: 10.1504/IJAISC.2022.126336

International Journal of Artificial Intelligence and Soft Computing, 2022 Vol.7 No.3, pp.179 - 193

Received: 19 Nov 2020
Accepted: 24 Sep 2021
Published online: 21 Oct 2022 *