GoldenFish Sentinel feature selection with SBM classifier for automatic seizure detection from EEG data

Among people who suffer from epilepsy, seizure detection is crucial to improving their wellbeing. Through the integration of sophisticated feature selection and a reliable classification algorithm, this study presents a new and effective way for seizure detection. To achieve precise and speedy seizure classification, the suggested method combines the GoldenFish Sentinel feature selection method with the Stoch Boundary Max (SBM) Classifier. The GoldenFish Sentinel method is modelled after fish's innate ability to choose the best characteristics from complex Electroencephalogram (EEG) inputs. The SBM Classifier is used because of its effective training and boundarymaximizing features. Significant benefits result from the combination of the SBM Classifier and GoldenFish Sentinel. In the beginning, the feature selection procedure cleans up the input data, making sure that only relevant features are used in the classification procedure. As a result, computational effectiveness and model resilience are increased. Second, by maximising the border between classes, the SBM Classifier is able to increase its discriminative power, which boosts seizure classification accuracy. The SBM Classifier's effective training capabilities enable real-time analysis of EEG inputs, making early seizure detection possible. This is crucial for swift intervention and enhanced patient outcomes. The proposed approach is rigorously evaluated on a comprehensive dataset of EEG recordings from individuals with epilepsy. Comparative analyses demonstrate its superiority over conventional methods in terms of accuracy and efficiency. By incorporating Stochastic Gradient Descent (SGD) and the SBM Classifier, the proposed approach effectively navigates local minima, facilitating efficient optimization in seizure detection with an accuracy of 98 %.