Home // eTELEMED 2024, The Sixteenth International Conference on eHealth, Telemedicine, and Social Medicine // View article
Authors:
Christoph Anders
Sai Siddhant Gadamsetti
Nico Steckhan
Bert Arnrich
Keywords: Mental Health; Mental Workload; Stress; Wearables; eHealth.
Abstract:
Prolonged levels of high mental workload and resulting stress are among the main causes of employee sickness. A possible solution would be implementing business rules based on objective analyses of stress levels and cognitive demands produced in employees by given tasks. This study laid the foundation for the development of personalized stress assistants. Physiological data of five groups of two participants were recorded, following a five-appointment study design. During the appointments, each pair underwent a cognitive load induction and subsequent stress reduction phase. Physiological signals were recorded with low-cost wearable sensors, subsequently analyzed for biomarkers, and compared for similarity between participants and groups. Results show that the sensors are capable of capturing descriptive data. Despite simultaneous task executions, it was found from the similarity analysis that the normalized Dynamic Time Warping distances between extracted features are greater for yoga sessions than during the cognitive load sessions. The classification of tasks was performed using the Machine Learning algorithms (i) Logistic Regression, (ii) Support Vector Machines, (iii) Nearest Neighbors, and (iv) Decision Trees trained on feature sets of either the Muse S, the Empatica E4, or both sensors together. Generalized as well as personalized models achieved classification accuracies over 85.00%. The recorded data is available upon request. The stimulus elicitation framework developed using PsychoPy and the software artifacts for data analysis were made publicly available, enabling the research community to evaluate their methods on this dataset and re-use analysis methods on their own or other datasets.
Pages: 16 to 24
Copyright: Copyright (c) IARIA, 2024
Publication date: May 26, 2024
Published in: conference
ISSN: 2308-4359
ISBN: 978-1-68558-167-1
Location: Barcelona, Spain
Dates: from May 26, 2024 to May 30, 2024