Publication

Abstract : Epilepsy is a persistent, non-communicable brain illness that affects around 50 million individuals worldwide. The repercussion of an epileptic seizure on the patient might be severe. People who encounter epileptic seizures typically find that a quick reaction from medical services is crucial since it can reduce the harmful effects. The Arduino Nano board is used in this design to help collect accelerometer data. Edge Impulse, an embedded machine-learning platform, receives the data. Edge Impulse processes this data and determines if the sample reflects a seizure or normal activity using Spectral characteristics and machine learning methods. The type and order of the filter are included in the block of spectral characteristics. Variations of high-pass and low-pass filters are compared to not employing any filters at all. The Dense layer neural network block aids in the model's ability to achieve better assessment metrics. The results show that a high pass filter used to block spectral features delivers the highest accuracy with a 96.3% rate.