ComEDA: A new tool for stress assessment based on electrodermal activity

Mimma Nardelli; Alberto Greco; Laura Sebastiani; Enzo Pasquale Scilingo

doi:10.1016/j.compbiomed.2022.106144

ComEDA: A new tool for stress assessment based on electrodermal activity

Comput Biol Med. 2022 Nov:150:106144. doi: 10.1016/j.compbiomed.2022.106144. Epub 2022 Sep 30.

Authors

Mimma Nardelli¹, Alberto Greco², Laura Sebastiani³, Enzo Pasquale Scilingo²

Affiliations

¹ Bioengineering and Robotics Research Centre E. Piaggio and Dipartimento di Ingegneria dell'Informazione, University of Pisa, Largo Lucio Lazzarino 1, Pisa, 56122, Italy. Electronic address: mimma.nardelli@unipi.it.
² Bioengineering and Robotics Research Centre E. Piaggio and Dipartimento di Ingegneria dell'Informazione, University of Pisa, Largo Lucio Lazzarino 1, Pisa, 56122, Italy.
³ Department of Translational Research and New Technologies in Medicine and Surgery, University of Pisa, via Paolo Savi 10, Pisa, 56126, Italy.

PMID: 36215850
DOI: 10.1016/j.compbiomed.2022.106144

Abstract

Non-specific sympathetic arousal responses to different stressful elicitations can be easily recognized from the analysis of physiological signals. However, neural patterns of sympathetic arousal during physical and mental fatigue are clearly not unitary. In the context of physiological monitoring through wearable and non-invasive devices, electrodermal activity (EDA) is the most effective and widely used marker of sympathetic activation. This study presents ComEDA, a novel approach for the characterization of complex dynamics of EDA. ComEDA overcomes the methodological limitations related to the application of nonlinear analysis to EDA dynamics, is not parameter-sensitive and is suitable for the analysis of ultra-short time series. We validated the proposed algorithm using synthetic series of white noise and 1/f noise, varying the number of samples from 50 to 5000. By applying our approach, we were able to discriminate a statistically significant increase of complexity in the 1/f noise with respect to white noise, obtaining p-values in the range [4.35 × 10^-6, 0.03] after the Mann-Whitney test. Then, we tested ComEDA on both EDA signal and its tonic and phasic components, acquired from healthy subjects during four experimental protocols: two inducing a sympathetic activation through physical efforts and two based on mentally stressful tasks. Results are encouraging and promising, outperforming state of the art metrics such as the Sample Entropy. ComEDA shows good performance not only in discriminating between stressful tasks and resting state (p-value < 0.01 after the Wilcoxon non-parametric statistical test applied to EDA signals of all the four datasets), but also in differentiating different trends of complexity of EDA dynamics when induced by physical and mental stressors. These findings suggest future applications to automatically detect and selectively identify threats due to overwhelming stress impacting both physical and mental health or in the field of telemedicine to monitor autonomic diseases correlated to atypical sympathetic activation. The Matlab code implementing the ComEDA algorithm is available online.

Keywords: Arousal; Complexity; Electrodermal activity (EDA); Phase space reconstruction; Skin conductance; Stress.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Algorithms*
Galvanic Skin Response*
Humans
Physical Examination