A Novel Feature Extraction Method for Classifying EEG Signals of two types of Reading: Normal Reading and Task-Specific Reading

Educational contents are designed without considering the students skills and interests. Usually, this scheme provokes a gap between the contents level and the students learning, which impacts in the emotional and stress level of the students. In this scenario, an automatic system to consider the special need of the students is required. For this reason, we develop a first step focused to implement a novel system called neurotutor, which would adapt the contents for each subject. Specifically, this step consists in assessing the differences in EEG signals recorded during two reading tasks: normal reading and task specific reading. Aiming to approach this issue, we propose a novel feature extraction method based on characteristics derived from CWT. This method outperformed the previous work and allowed save computational time consumptions. As a future study, we must investigate if these outcomes can be kept increasing the number of subjects, and considering additional tasks involved during learning processes. Clinical Relevance — A first step targeted to implement an EEG-based neurotutor is presented. This automatic system would help to reduce the student's stress during the process of learning.