Title: Discrete wavelet transform analysis and empirical mode decomposition of physiological signals for stress recognition
Authors: Khadidja Gouizi; Fethi Bereksi Reguig; Choubeila Maaoui
Addresses: Biomedical Engineering Research Laboratory, University Abou Bakr Belkaid, 22 Abi Ayad Abdelkrim street, Fg Pasteur, 13000, Tlemcen, Algeria ' Biomedical Engineering Research Laboratory, University Abou Bakr Belkaid, 22 Abi Ayad Abdelkrim street, Fg Pasteur, 13000, Tlemcen, Algeria ' Conception, Optimisation and Modeling Systems Laboratory, University of Lorraine, 7 Marconi street, 57070, Metz, France
Abstract: Stress is universally known to be a contributing factor in developing of many diseases. This work focuses on developing a user-independent and user-dependent stress recognition system using five physiological signals that are electromyogram, galvanic skin response, skin temperature, blood volume pulse and respiratory response. Emotional data is collected from 33 subjects by using Stroop game. These are processed using Discrete Wavelet Transform (DWT) and Empirical Mode Decomposition (EMD). Also, decomposition of physiological signals into intrinsic mode functions (IMFs) by the EMD method is done. With the Support Vector Machine (SVM), the classification results show a better accuracy using the DWT method compared to those of the EMD method. For the user-independent study an overall classification accuracy of 60.9% in stress recognition is reached whereas for the user-dependent study an overall classification accuracy of 80% is achieved. In addition, overall recognition rates attain 100% when using the DWT.
Keywords: physiological signals; discrete wavelet transform; empirical mode decomposition; stress recognition; support vector machine.
International Journal of Biomedical Engineering and Technology, 2018 Vol.27 No.3, pp.247 - 270
Received: 12 May 2016
Accepted: 26 Aug 2016
Published online: 06 Aug 2018 *