Detection of seizures from EEG using KNN and Opti-KNN

Mohebbanaaz Mohebbanaaz (Ravindra College of Engineering for Women)
Madhusudhana Reddy Barusu (G. Pullaiah College of Engineering and Technology)
S. Bhargav (VNR Vignana Jyothi Institute of Engineering and Technology)
D. Lalitha Kumari (JNTUACEA)
Vishnu Vardhan Rao G. (Vel Tech Multi Tech Dr. Rangarajan Dr. Sakunthala Engineering College)
M. Jyothirmai (Ravindra College of Engineering for Women)

International Journal of Informatics and Communication Technology (IJ-ICT)
Vol 15, No 4: December 2026

Epileptic seizures are caused by sudden and abnormal electrical disturbances in the brain, which can be identified through electroencephalogram (EEG) signals. Detecting seizures accurately at an early stage is important for providing timely treatment and improving patient care. In this study, two machine learning (ML) methods, namely k-nearest neighbour (KNN) and an optimized version of KNN (Opti-KNN), are proposed for automated seizure detection using EEG data. The research was carried using the BONN EEG dataset. The standard KNN model produced an accuracy of 94%, whereas the optimized KNN model improved the performance and achieved 97.33% accuracy. The findings indicate that proper hyperparameter tuning can greatly enhance the effectiveness of seizure classification. Overall, the proposed models provide a consistent and effective solution for automated seizure detection.