EEG Seizure Alarm System
EWH Design Competition 2021 Third Place
Overview
Objective
Submit an innovative design for medical technology in low-resource settings
Propose a seizure prediction system consisting of a neural network and an associated low-cost wearable alarm system
Process
Communicated with Open BCI to acquire their Ultracortex Mark IV model with Cyton Board
Modified the CAD model of the headset with Fusion 360 to implement an elastic design that would lower the impedance between the brain and electrodes
Filtered the raw EEG signals with the Open BCI Cyton+Daisy board which will be sent to a Raspberry Pi via Bluetooth
Predicts the onset of a seizure by detecting the presence of preictal waveforms with a machine-learning model in the Raspberry Pi
Result
The headset was successful at fitting around the user’s head, which allowed successful detection and filtering of EEG signals
With the seizure machine learning algorithm, it was able to predict seizures of testing data and alert the user before seizures occurred