Gated Recurrent Unit Network for Psychological Stress Classification Using Electrocardiograms from Wearable Devices

Area of Science:

• Biomedical Engineering
• Psychophysiology
• Wearable Technology

Background:

• Wearable devices are increasingly used for psychological stress research.
• Individual differences and data collection costs pose challenges.
• Accurate stress detection from physiological signals is needed.

Purpose of the Study:

• To develop a model for detecting psychological stress states using electrocardiogram (ECG) signals.
• To address challenges in individual variability and data collection costs.
• To build a neural network model for real-time stress monitoring.

Main Methods:

• A virtual reality (VR) high-altitude experiment was designed to induce psychological stress.
• Participants wore smart ECG T-shirts to collect synchronized ECG data.
• A deep gated recurrent unit (GRU) neural network analyzed heart rate variability (HRV) features.

Main Results:

• The GRU model achieved superior classification performance across four stress states: resting, VR scene adaptation, VR task, and recovery.
• The developed method demonstrated effectiveness in capturing temporal stress variations.
• The model outperformed existing comparison methods in stress detection accuracy.

Conclusions:

• The proposed ECG-based GRU model is effective for psychological stress detection.
• This approach offers a viable remote stress monitoring solution for specialized industries.
• Further research can leverage this model for broader applications in mental health monitoring.