Development of an Embedded Electroencephalography System for Epilepsy Seizure Detection in Community Health Settings in Nigeria

Epilepsy is highly prevalent in Nigeria and severely underdiagnosed and undertreated, particularly in communities remote from tertiary neurology services, and development of affordable portable EEG systems for seizure detection in community health settings addresses a critical diagnostic gap. This study develops and validates an embedded four-channel electroencephalography system for epilepsy seizure detection in community health settings in Nigeria. The hardware employs ADS1299 ultra-low-noise EEG-specific analogue front-end integrated circuits with hardware programmable gain amplification up to 24, passive high-pass filtering at 0.5 Hz, and a 250 Hz sampling rate, managed by a Raspberry Pi Zero W for signal processing and wireless data transmission. A signal processing and seizure detection pipeline was implemented combining bandpass filtering, independent component analysis for ocular and muscular artifact rejection, and a support vector machine classifier trained on a combined clinical EEG dataset supplemented by recordings from forty epilepsy patients at the Federal Medical Centre Makurdi, Benue State. Classifier training used time-frequency features extracted from continuous wavelet transform decomposition. System validation on a held-out test set from the Makurdi patient dataset achieved seizure detection sensitivity of 87.2 percent and false detection rate of 0.8 per hour during interictal periods. System power consumption confirmed eight hours of battery operation. Signal quality on ten community volunteers wearing dry electrodes confirmed adequate impedance and noise floor for seizure detection application. The study recommends clinical validation in a community EEG screening setting and integration of a teleconsultation pathway for remote neurologist review. Keywords: EEG, seizure detection, epilepsy, embedded system, community health settings.