EEG Conformer Model Based Epileptic Seizure Prediction Using Deep Learning

Epilepsy is a disorder of the brain characterized by repeated seizures. A seizure is usually defined as a sudden alteration of behavior due to a temporary change in the electrical functioning of the brain. The emergence of deep learning methodologies has incentivized inventive applications within the domain of medical diagnostics, specifically concerning the prediction of epileptic seizures. In the existing method SVM classifiers was used to differentiate between seizure and non-seizure states based on the extracted features, providing a well-established machine learning framework for binary classification tasks. Used ensemble learning strategies, such as Random Forests and Decision Trees, to combine several models and identify intricate links in the EEG data, improving classification accuracy and accuracy is upto 89.7%. The convolutional Transformer Model, the deep learning architecture designed for precise and timely prediction of epileptic seizures, is proposed in this project. By utilizing the complex patterns and intrinsic focal relationships found in EEG signals, the model is able to extract subtle information that is essential for differentiating between preictal states. Compared to other approaches, the EEG Conformer Model performs better at predicting seizures after training on a large dataset of CHB MIT EEG database. By adding conformer blocks, the model is better to identify long-range dependencies in EEG sequences, which increases prediction accuracy. This project focuses on the development and implementation of an EEG conformer model for epileptic seizure prediction, employing deep learning methodologies. By training the model on diverse and extensive datasets, it is expected to learn and generalize underlying patterns, enabling it to predict seizures with a high degree of accuracy.