What is the clinical evidence from this study?

Study design: Observational. Population: Acute stress (n=65). Intervention: Electrodermal activity (EDA) modeling and SVM classification vs. Baseline (non-stress) condition. Primary outcome: Accuracy of binary stress vs. non-stress classification.

January 29, 2021Open Access

Acute Stress State Classification Based on Electrodermal Activity Modeling

Key Result

An electrodermal activity-based modeling pipeline using a support vector machine successfully detected acute stress conditions with an average accuracy of 94.62%.

Study Design

Type

Observational (n=65)

Structured PICO

Population

65 healthy young adults (mean age 21.8 years) subjected to a modified Trier Social Stress Test to evaluate an electrodermal activity-based stress classification model.

Exposure

Methodological pipeline combining cvxEDA for electrodermal activity (EDA) processing and decomposition with a support vector machine (SVM) algorithm for stress state classification.

Comparator

Non-stress condition (baseline stage)

Outcome

Accuracy of stress detection (binary stress vs. non-stress, and 4-class pattern recognition)

A novel pipeline using electrodermal activity and support vector machine modeling can accurately detect and classify acute stress states in controlled settings.

Limitations

Accuracy was estimated on data acquired in controlled laboratory conditions, which may not reflect real-life scenarios with confounding factors.
The fixed-order stimulation of the TSST design might affect the perception of stress induced at each stage.
The sample population was very homogeneous in age, limiting the extrapolation of results to other age groups.
The study did not investigate the relationship between physiological (sympathetic) reactions and subjective-psychological (affective) responses.

Abstract

Acute stress is a physiological condition that may induce several neural dysfunctions with a significant impact on life quality. Accordingly, it would be important to monitor stress in everyday life unobtrusively and inexpensively. In this paper, we presented a new methodological pipeline to recognize acute stress conditions using electrodermal activity (EDA) exclusively. Particularly, we combined a rigorous and robust model (cvxEDA) for EDA processing and decomposition, with an algorithm based on a support vector machine to classify the stress state at a single-subject level. Indeed, our method, based on a single sensor, is robust to noise, applies a rigorous phasic decomposition, and implements an unbiased multiclass classification. To this end, we analyzed the EDA of 65 volunteers subjected to different acute stress stimuli induced by a modified version of the Trier Social Stress Test. Our results show that stress is successfully detected with an average accuracy of 94.62 percent. Besides, we proposed a further 4-class pattern recognition system able to distinguish between non-stress condition and three different stressful stimuli achieving an average accuracy as high as 75.00 percent. These results, obtained under controlled conditions, are the first step towards applications in ecological scenarios.

Bookmark

View Full Paper