EEG, EMG and ECG based Determination of Psychosocial Risk Levels in Teachers based on Wavelet Extreme Learning Machine Autoencoders

Abstract

Teachers who perform a sacred work are faced with many psychosocial risks. These risks can often be caused by the school administration, the students, and environmental factors. Machine learning and data mining approaches have recently gained much attention in social and educational researches. In this study, a novel approach, which is based on data augmentation and data classification, is proposed for the prediction of the psychosocial risk levels of the teachers. The data augmentation is carried out by using an extreme learning machine autoencoders (ELM-AE). More specifically, the wavelet activation function is incorporated into the ELM-AE to develop a novel approach called WELM-AE. After data augmentation, a traditional ELM classifier is used in the prediction of the psychosocial risk levels of teachers. A dataset, which contains physiological factors, namely Electrocardiography (ECG), Electromyography (EMG), and Electroencephalography (EEG), is used to evaluate the performance of the proposed method. Classification accuracy is used as the evaluation metric. All coding is carried out in MATLAB, and a 99.9% accuracy score is obtained with the proposed method. A performance comparison is also carried out with some machine learning techniques, namely decision trees (DT), support vector machines (SVM), and K-nearest neighbour (KNN). The results show that the proposed WELM-AE and ELM classifier outperform the compared methods.