Web Application-Based Epileptic Seizure Prediction using Machine Learning Approach

Epileptic seizures are unpredictable neurological events of temporary disabilities in movement, behavior, sensation, and consciousness produced by abnormal and excessive electrical activities in the brain. Accurately anticipating epileptic seizures in their early phases is crucial for treatment initiation and patient safety. Traditional seizure prediction models struggle with high-dimensional EEG data, leading to suboptimal accuracy and inefficiency of processing. Current approaches are not easily deployable, which makes them inaccessible for practical settings. The Random Forest Classifier (RFC)-based Epileptic Seizure Prediction Model presented in this research is incorporated into a web application for user-friendly disease diagnosis. The model reduces dimensionality and maintains important EEG signal patterns using Principal Component Analysis (PCA) to select features. The Random Forest classifier improves interpretability and robustness while surpassing traditional models in generalization and processing efficiency. Patients can enter EEG data values through the web interface and receive real-time seizure likelihood estimates. This method increases the precision of predictions and generates a user-friendly platform for proactive seizure control. The evaluation demonstrated high precision and recall in detecting the seizure patterns on the benchmark EEG datasets.