Utility of the Full ECG Waveform for Stress Classification.

Area of Science:

• Biomedical Engineering
• Psychophysiology
• Signal Processing

Background:

• Psychological stress detection commonly uses R-peak derived heart rate variability (HRV) from electrocardiogram (ECG) signals.
• Smaller ECG waves (P, Q, S, T) offer potential HRV features but are susceptible to noise, hindering reliable detection.

Purpose of the Study:

• To investigate the benefits of incorporating HRV features from smaller ECG waves (P, Q, S, T) into stress detection algorithms.
• To analyze the performance and reliability of an automated peak detection algorithm for these smaller ECG waves.

Main Methods:

• Utilized a dataset of ECG recordings from 57 participants experiencing spider phobia during video exposure.
• Compared two binary stress classification models: one using only R-peak features, the other including novel small-peak features.
• Analyzed model performance using precision, recall, F1-score, AUC, accuracy, and Akaike Information Criterion (AIC).

Main Results:

• While standard performance metrics were similar, the model with small-peak features was significantly more probable to minimize information loss (AIC).
• Small-peak features demonstrated stronger regression coefficients, indicating a more robust relationship with acute psychological stress than R-peak features.
• Automated detection reliability analysis was performed for each smaller ECG wave component.

Conclusions:

• Novel small-peak ECG intervals may reflect independent cardiac processes indicative of psychophysiological stress responses.
• Integrating these novel features into stress detection algorithms holds potential for improved accuracy and deeper understanding of stress.
• Further research into small-peak interval analysis could unlock new avenues for non-invasive stress monitoring.