Machine Learning-Based Epileptic Seizure Detection using EEG Data in Zimbabwe

Epilepsy is a critical neurological disorder, especially prevalent in developing nations like Zimbabwe, where healthcare infrastructure and diagnostic resources remain limited. Electroencephalograms (EEGs) provide essential clinical insight into seizure activity but require expert interpretation, often leading to delays in diagnosis. This research presents a comparative study of three machine learning models—Random Forest (RF), K-Nearest Neighbours (KNN), and Convolutional Neural Networks (CNN)—for the automated detection of epileptic seizures using a Zimbabwe-specific EEG dataset. The dataset comprises 2,216 EEG segments, each with 667 extracted features. The study investigates each model's effectiveness using metrics such as accuracy, precision, recall, F1-score, and ROC-AUC. Among all models, RF achieved the highest classification performance with an accuracy of 85.78%, suggesting strong potential for integration into clinical decision-support tools to aid early and reliable epilepsy diagnosis in under-resourced settings.