EEG Based Deep Learning Model for Automated Classification of Frequent and Infrequent Readers

Reading comprehension is a complex cognitive process that involves semantic decoding, and schema integration influenced by reading practice. This study investigates the neural correlates of reading comprehension and explores how brain activation patterns differ between frequent and infrequent readers. Using Electroencephalography (EEG), the brain activation patterns were recorded from participants while performing the reading comprehension tasks. The EEG signal is recorded with a sampling rate of 1024 Hz to capture the dynamic changes in brain activation during reading tasks, revealing how frequent and infrequent readers differ in their cognitive processing and neural efficiency. The 21 channels of EEG signal acquired from 9 participants were pre-processed to obtain a 512 Hz band-limited signal. The pre-processed EEG signals were segmented before being fed into a twenty-four-layered convolutional neural network (CNN) model. The convolution stage extracts the temporal and spatial features, and the most significant features are extracted at the max-pooling stage. Frequent and infrequent readers are classified using the fully connected layer based on their EEG signals. The proposed model obtained a classification accuracy of 93.94%.