Real-Time Detection of Epileptiform Activity and Seizure Onset Using EEG Signals and Machine Learning

Biomedical Engineering

Dhyeaya Dhiren Parmar

Abstract

Epilepsy affects 50 million people worldwide, and most seizures happen without warning, leaving patients at risk of serious injury. While researchers have tried to predict seizures using brain wave patterns, existing methods struggle to work across different patients.

I developed a machine learning system that analyzes EEG signals and achieved 87.5% accuracy in detecting pre-seizure brain activity across multiple patients, with a 2-minute warning window.

Video

https://youtu.be/vXclnmwKAt8

Research poster

Faculty mentor

Portrait of Aurel Coza

Aurel Coza

Center Director and Professor of Practice

Corporate Engagement & Strategic Partnerships

[email protected]