Songklanakarin Journal of Science and Technology (SJST) (Dec 2021)
Application of Hjorth parameters in the classification of healthy aging EEG signals
Abstract
Aging has extensive impacts on brain cognition. In this work we proposed a method using Hjorth parameters to classify the elderly’s electroencephalography (EEG) signals from that of middle age group by applying K-nearest neighbor (KNN) and Random forest (RF) classifiers. We acquired EEG of 20 healthy middle age subjects and 20 healthy elderly subjects in resting state eyes-open for 5 minutes and eyes-closed for 5 minutes using an 8-electrodes device. Euclidean and Manhattan distance measures were tested using KNN. The classifier performance was evaluated by using accuracy, sensitivity, specificity, and kappa statistic. The best accuracy achieved was 91.25 %, and kappa statistic of 0.825, in eyes-closed state. In eyes-open state 90% accuracy was achieved with kappa statistic of 0.80. RF achieved 83.75% accuracy with kappa statistic of 0.675 in eyes-closed state and 78.75% accuracy with Kappa statistic of 0.575 in eyes-open state. The KNN performed better using Manhattan distance function in both eyes-open and eyes-closed states. Results showed the potential of Hjorth parameters as the suitable EEG features in the classification of EEG aging signals.
Keywords