Integrated TSVM-TSK fusion for enhanced EEG-based epileptic seizure detection: Robust classifier with competitive learning

Early diagnosis of epilepsy is crucial for patient survival and well-being, making it essential to develop effective methods for early disease detection based on health parameters. This paper introduces a novel approach to enhance electroencephalogram (EEG)-based epileptic seizure detection, presenting a robust Takagi-Sugeno-Kang (TSK) fuzzy classifier. Our key contribution lies in the integration of advanced techniques, specifically, possibility-based clustering with competitive learning (PC-CL), into the classifier. This not only refines fuzzy rule parameters with unprecedented precision but also imparts resilience against noise, elevating the interpretability of the model. The distinctive amalgamation of PC-CL ensures a more nuanced and accurate representation of complex EEG patterns associated with epileptic seizures. Additionally, a hybrid approach combining a twin support vector machine with a fuzzy system (TSVM-PC-CL-TSK) is proposed for epilepsy classification. Results from experiments show that the proposed classifier performs more efficiently than previous, achieving an impressive 99.41 % accuracy rate in identifying abnormal EEG signals associated with epileptic seizures. The research showcases the significant potential of this innovative approach in epilepsy research and clinical applications, offering improved epileptic seizure detection and elevated patient care standards.