Fig. 5

Results of subject-wise analysis: a Classification performance of different features with respect to different p value thresholds that used in feature selection method. Average classification accuracy of time–frequency power spectrum features of all 64 electrodes (TFPS64), left hemispheric electrodes (TFPSL), right hemispheric electrodes (TFPSR) and differential asymmetry between hemispheric features (DATFPS) are represented along with empirical chance level (pink horizontal line). Error bars indicate standard error of mean (SEM). b Representation of number of selected features and average classification accuracy of DATFPS feature with respect to different p value thresholds as DATFPS feature set yielded the best performance for all subjects. c Sensitivity and specificity performance (in %) for each feature type. Error bars indicate standard deviation (SE) across subjects. d Representation of occurrence count of dominant features. Band-wise dominant features for each subject is shown for DATFPS feature type. Among five EEG frequency bands, maximum selected features belonged from alpha frequency band. e Temporal course of occurrence count of dominant features. Error bars indicate SEM across subjects