Classification of Working Memory Impairment in Children Using Electroencephalograph Signal at the Prefrontal Cortex

This study emphases on the alpha oscillation of Electroencephalography (EEG) signal of normal children ability towards working memory performance and visual responsive. The assessments were conducted on 30 children aged between 7 to 9 years old who have no records of working memory disability. The raw EEG signals were decomposed using discrete wavelet transform with mother wavelet: Daubechies 4 (db4) at a level of decomposition of 8. The EEG signals were recorded at the prefrontal cortex (channel F3, F4, FZ, F7, and F8) during two visual assessments. The feature extractions at alpha frequency are mean and standard deviation of voltage activity during brain activation. Support Vector Machine (SVM) is used to identify when the working memory performance and visual sensory interaction occur. The percentage accuracy for Phase 1 is 73.33 % for the control group and 26.67 % for the test group. However, at the Phase 2, the percentage of accuracy for the control group decreased to 56.67 % and the accuracy for the test group is 43.33 %. Results show that the accuracy of working memory is affected by the task given. In conclusion, the study proposed that age and the increased level of tasks affect working memory performance.