Academic stress detection on university students during COVID-19 outbreak by using an electronic nose and the galvanic skin response

• Published in: Biomedical signal processing and control Vol. 68; p. 102756
• Main Authors: Durán Acevedo, Cristhian Manuel, Carrillo Gómez, Jeniffer Katerine, Albarracín Rojas, Camilo Andrés
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.07.2021
• Subjects: Academic stress; Classification methods; Electronic nose; Galvanic skin response; SISCO inventory; SISCO inventory; Academic stress; Galvanic skin response; Classification methods; Electronic nose
• ISSN: 1746-8094; 1746-8108; 1746-8094
• DOI: 10.1016/j.bspc.2021.102756

[Display omitted] •New non-invasive method for academic stress detection by using an electronic nose system.•Implementation of a stress detector based on volatile organic compounds emitted by the skin.•The academic stress level during Covid-19 is classified with different algorithms.•A 96 % success rate is achieved with the E-nose signals and 100 % with the GSR signals. Academic stress is an emotion that students experience during their time at the university, sometimes causing physical and mental health effects. Because of the COVID-19 pandemic, universities worldwide have left the classroom to provide the method of teaching virtually, generating challenges, adaptations, and more stress in students. In this pilot study, a methodology for academic stress detection in engineering students at the University of Pamplona (Colombia) is proposed by developing and implementing an artificial electronic nose system and the galvanic skin response. For the study, the student’s stress state and characteristics were taken into account to make the data analysis where a set of measurements were acquired when the students were presenting a virtual exam. Likewise, for the non-stress state, a set of measurements were obtained in a relaxation state after the exam date. To carry out the pre-processing and data processing from the measurements obtained previously by both systems, a set of algorithms developed in Python software were used to perform the data analysis. Linear Discriminant Analysis (LDA), K-Nearest Neighbors (K-NN), and Support Vector Machine (SVM) classification methods were applied for the data classification, where a 96 % success rate of classification was obtained with the E-nose, and 100 % classification was achieved by using the Galvanic Skin Response.