Epilepsy predictor

Embedded Biomedical System for Epileptic Seizure Prediction

Our project was inspired by the challenges faced by people living with epilepsy, where unexpected seizures can cause fear, injury, and continuous stress for patients and caregivers. Since conventional systems generally detect seizures after they begin, we aimed to develop a non-invasive wearable system capable of providing an early warning before seizure onset.

We proposed a smart headband/cap integrated with EEG and motion/physiological sensors to continuously monitor relevant biomedical signals. The acquired signals are amplified, filtered, digitized, and analyzed using Artificial Intelligence to identify abnormal patterns associated with the preictal phase. Machine-learning and deep-learning approaches such as SVM, Random Forest, KNN, CNN, RNN, and LSTM/GRU were considered for pattern recognition and prediction. When a potential seizure is identified, the system is designed to generate an alert and notify caregivers.

A major challenge was handling noise and motion artifacts while maintaining reliable continuous monitoring and low power consumption. Through this project, we gained practical knowledge in EEG signal processing, wearable biomedical systems, embedded technology, and AI-based healthcare applications. We are proud of developing an integrated concept that brings together biomedical engineering and AI toward earlier, safer, and more personalized seizure management.