EEG-Based Cognitive Impairment Classification using Support Vector Machine Algorithm

Electroencephalography (EEG) offers a non-invasive means to evaluate brain activity by recording brain signals, commonly employed in assessing cognitive functions and related dysfunctions triggered by external stimuli such as light, sound, and cognitive tasks. This study aims to detect cognitive impairment (CI) using resting-state EEG data obtained from 70 participants via 16-channel EEG recordings. Given that CI correlates with both physiological and psychological alterations in the brain, relevant features are extracted using algorithms like phase locking value (PLV) and band power. For binary classification, a support vector machine (SVM) algorithm is created that uses a radial basis function (RBF) kernel. The study’s outcomes reveal that the classifier can discern EEG signals with an accuracy of 85.71%. In optimizing the feature set for the tasks at hand, 12 EEG channels, which exhibit relatively higher activity, are selected, leading to a 25% reduction in the feature set. Despite this reduction, SVM maintains the same classification accuracy of 85.71%. Notably, the proposed method demonstrates superior accuracy compared to contemporary research efforts in the field.