TY - GEN
T1 - Evaluating a New Approach to Data Fusion in Wearable Physiological Sensors for Stress Monitoring
AU - Rodrigues, Clarissa
AU - Fröhlich, William R.
AU - Jabroski, Amanda G.
AU - Rigo, Sandro J.
AU - Rodrigues, Andreia
AU - de Castro, Elisa Kern
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - The physiological signs are a reliable source to identify stress states, and wearable sensors provide precise identification of physiological signs associated with the stress occurrence. The literature review shows that the use of physiological signs as a source for stress patterns identification is still a critical investigation subject. Few studies evaluate the effect of combining several different signals and the implications of the data acquisition procedures and details. This article’s objective is to investigate the possible integration of data obtained from heart rate variability, electrocardiographic, electrodermal activity, and electromyography to detect stress patterns, considering a new experimental protocol to data acquisition. The data acquisition involved the Trier Social Stress Test, wearable sensor monitoring, and complementary stress perception instruments, resulting in a publicly available dataset. This dataset was evaluated using different machine learning classifiers, considering the obtained annotated data and exploring different physiological features and their combinations.
AB - The physiological signs are a reliable source to identify stress states, and wearable sensors provide precise identification of physiological signs associated with the stress occurrence. The literature review shows that the use of physiological signs as a source for stress patterns identification is still a critical investigation subject. Few studies evaluate the effect of combining several different signals and the implications of the data acquisition procedures and details. This article’s objective is to investigate the possible integration of data obtained from heart rate variability, electrocardiographic, electrodermal activity, and electromyography to detect stress patterns, considering a new experimental protocol to data acquisition. The data acquisition involved the Trier Social Stress Test, wearable sensor monitoring, and complementary stress perception instruments, resulting in a publicly available dataset. This dataset was evaluated using different machine learning classifiers, considering the obtained annotated data and exploring different physiological features and their combinations.
KW - Machine learning
KW - Stress
KW - Wearable sensors
UR - http://www.scopus.com/inward/record.url?scp=85094133170&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-61380-8_37
DO - 10.1007/978-3-030-61380-8_37
M3 - Conference contribution
AN - SCOPUS:85094133170
SN - 9783030613792
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 544
EP - 557
BT - Intelligent Systems - 9th Brazilian Conference, BRACIS 2020, Proceedings
A2 - Cerri, Ricardo
A2 - Prati, Ronaldo C.
PB - Springer Science and Business Media Deutschland GmbH
T2 - 9th Brazilian Conference on Intelligent Systems, BRACIS 2020
Y2 - 20 October 2020 through 23 October 2020
ER -