A Multi-feature Fuzzy Index to Assess Stress Level from Bio-signals.

Area of Science:

• Biomedical Engineering
• Physiological Computing
• Affective Computing

Background:

• Accurate stress detection is crucial for monitoring well-being and performance.
• Existing multi-feature stress indices often require complex tuning and feature selection.
• A simplified approach using single physiological features could enhance stress assessment.

Purpose of the Study:

• To introduce a novel mono-feature fuzzy index for evaluating stress levels.
• To assess the index's performance in classifying stress and non-stress situations.
• To explore the index's utility in feature selection and stress detection.

Main Methods:

• Development of a fuzzy index based on a single feature from electrocardiogram (ECG) or galvanic skin response (GSR) signals.
• Training the index using feature measures recorded during resting states.
• Analysis of the index's performance on a dataset of 160 time periods from 20 subjects performing stressful and control tasks.

Main Results:

• The mono-feature fuzzy index demonstrated the ability to be integrated into multi-feature systems without recalibration.
• The developed stress index achieved 72% accuracy in correctly classifying stress versus no-stress periods.
• The study provided insights into the effectiveness of different tasks in inducing physiological stress.

Conclusions:

• A single physiological feature, when processed through a fuzzy index, can effectively detect stress.
• The proposed mono-feature fuzzy index offers a computationally efficient and tunable component for advanced stress monitoring systems.
• This approach facilitates feature selection and stress detection, contributing to the field of affective computing.