HRV Features as Viable Physiological Markers for Stress Detection Using Wearable Devices.

Area of Science:

• Physiological monitoring
• Machine learning applications
• Automotive safety

Background:

• Stress is a significant factor in automobile accidents, leading to injuries and fatalities.
• Physiological measurements, particularly heart rate variability (HRV), offer a quantifiable method for stress assessment.
• Wearable devices provide accessible platforms for continuous physiological data collection.

Purpose of the Study:

• To investigate the efficacy of HRV-derived features as reliable stress markers.
• To develop and compare machine learning models for accurate stress level classification.
• To explore the potential of using wearable device data for non-invasive stress detection.

Main Methods:

• Extracted HRV features from ECG data of automobile drivers.
• Developed and evaluated machine learning models including KNN, SVM, MLP, RF, and GB.
• Validated models for stress classification accuracy.
• Focused on key HRV metrics like AVNN, SDNN, and RMSSD.

Main Results:

• HRV features demonstrated strong potential as stress indicators.
• The best-performing machine learning model achieved an 80% recall rate in stress detection.
• Specific HRV metrics (AVNN, SDNN, RMSSD) were identified as crucial for stress detection.

Conclusions:

• HRV analysis, combined with machine learning, offers a viable method for non-invasive stress detection.
• The developed models can be adapted for stress monitoring in diverse applications.
• This approach supports applications in physical rehabilitation, anxiety management, and mental wellbeing.