Measuring Heart Rate Variability Using Facial Video

Area of Science:

• Biomedical Engineering
• Cardiovascular Health
• Signal Processing

Background:

• Heart Rate Variability (HRV) is crucial for assessing cardiac health and is typically measured via electrocardiogram (ECG).
• Photoplethysmography (PPG) offers an alternative, but video-based measurement presents a more accessible, non-invasive, and hands-free approach.

Purpose of the Study:

• To develop and validate a novel methodology for extracting HRV from video recordings.
• To compare the accuracy and reliability of video-derived HRV metrics against traditional PPG measurements.

Main Methods:

• A new methodology was developed using face detection algorithms and color augmentation to extract HRV from video.
• The methodology was applied to 45 samples, with simultaneous PPG recordings for comparison.
• Extracted signals were analyzed using time-domain, frequency-domain, and non-linear methods, with statistical validation including Pearson, Spearman, and Kendall correlations and Benjamini-Hochberg correction.

Main Results:

• Video and PPG measurements of heart rate showed a mean error of less than 1 bpm across all subjects.
• Out of 61 computed HRV variables, 38 demonstrated significant correlations between video and PPG methods.
• Key correlated variables included Heart Rate (0.991), Mean NN Interval (0.990), and NN Interval Count (0.955).

Conclusions:

• Video-based HRV measurement is a viable, accurate, and accessible alternative to PPG.
• The developed methodology effectively extracts clinically relevant HRV parameters from video data.
• This non-invasive technique holds promise for widespread application in remote cardiac health monitoring and risk assessment.