Automated Epilepsy Seizure Detection in Pediatric Using Machine Learning

Abstract

Epilepsy is a prevalent neurological disorder marked by recurrent and unpredictable seizures. In pediatric patients, early and accurate detection is critical to enabling timely medical intervention and improving long-term health outcomes. This thesis presents the development of an automated system for epileptic seizure detection in children using machine learning techniques. The proposed approach leverages a one-dimensional convolutional neural network (1D-CNN) model to analyse and classify CHB-MIT EEG data for the detection of ictal events. The system demonstrates high performance, achieving an accuracy of 97%, sensitivity of 97.03% and specificity of 96.83%. These results indicate the model’s strong ability to distinguish between ictal and preictal states, with a low false positive rate (3.17%) and false negative rate (2.97%). The results are promising and highlight the potential of the proposed system in supporting pediatric seizure detection.