DOI: https://doi.org/10.59321/BAUETJ.V4I2.15
AUTHOR(S)
Md. Mushfikur Rahman Khan1*, Md. Nazmul Islam1, Md. Mahmudul Hasan Rashed1, Afzal Hossen1
ABSTRACT
In this research, a method for automatically detecting mental arithmetic problems and evaluating their difficulty level is presented. This method uses single-channel EEG signals. It is essential to comprehend the effects of varying task difficulty on the cerebral cortices and to quantitatively evaluate the operation of different brain waves during cognitive tasks. A filter bank divides the EEG data into different rhythms (gamma, beta, alpha, theta, and delta). Following that, each rhythm is assessed according to several factors, including energy, entropy, mean, L2 norms, skewness, kurtosis, relative power, and absolute power. Metrics like precision, recall, F1 score, and confusion matrix are computed using the SVM classifier, a machine learning classifier. The program effectively distinguishes between the brain’s active and resting states during the specified cognitive task. The maximum accuracy of 86.67% was obtained while employing 4 characteristics for subject F8 and the delta sub-band. The maximum accuracy of 86.67% was obtained for subject Fp1 and the beta sub-band when 8 characteristics were used.