Comparison of Stress Detection through ECG and PPG signals using a Random Forest-based Algorithm.

Area of Science:

• Affective computing
• Neuroscience
• Psychology
• Medicine

Background:

• Stress is a major health concern in the 21st century, with ongoing research across multiple scientific disciplines.
• Current stress detection relies on cortisol levels, lacking a standard for continuous monitoring.
• Wearable sensors offer potential for stress identification via physiological signals like Heart Rate Variability (HRV).

Purpose of the Study:

• To develop and validate a supervised machine learning algorithm for stress detection using HRV.
• To investigate the efficacy of photoplethysmograms (PPG) as a low-cost alternative to electrocardiograms (ECG) for HRV analysis and stress detection.

Main Methods:

• A subject-independent Random Forest algorithm was trained and tested using HRV features derived from ECG and PPG signals.
• Data was collected from 46 healthy subjects.
• The algorithm was evaluated for its accuracy in stress identification.

Main Results:

• The machine learning algorithm achieved over 80% F1-score and 90% AUC for stress detection from both ECG and PPG data.
• Photoplethysmography (PPG) demonstrated comparable performance to electrocardiography (ECG) in HRV analysis and stress detection.
• The proposed method enables accurate, automated, and continuous stress analysis.

Conclusions:

• Photoplethysmography (PPG) is a suitable and cost-effective alternative to ECG for Heart Rate Variability (HRV) analysis.
• The developed machine learning algorithm effectively detects stress using HRV from PPG signals.
• This technology has significant potential for researchers and clinicians in automated, continuous stress monitoring.