Data Reduction Methodology for Dynamic Characteristic Extraction in Photoplethysmogram

Area of Science:

• Biomedical Engineering
• Physiological Signal Processing
• Nonlinear Dynamics

Background:

• Photoplethysmogram (PPG) signals are vital for non-invasive health monitoring in wearables.
• Green-light PPG (gPPG) shows potential for improved signal quality over red-light PPG (rPPG).
• Nonlinear time series analysis extracts complex health information from PPG signals.

Purpose of the Study:

• To investigate the impact of downsampling frequencies on PPG signal analysis.
• To determine the minimum time series length for accurate nonlinear analysis of PPG signals.
• To compare the effectiveness of gPPG and rPPG for dynamical characteristic estimation.

Main Methods:

• Examined downsampling frequencies and varying time series lengths for gPPG and rPPG signals.
• Assessed the accuracy of estimating dynamical characteristics under different data conditions.
• Performed comparative analysis between gPPG and rPPG signal performance.

Main Results:

• A sampling frequency of 200 Hz offers an optimal balance between signal accuracy and computational load.
• Dynamical properties of PPG signals stabilize sufficiently around 170 seconds, with less than 5% error.
• No significant statistical differences were found between gPPG and rPPG in estimating dynamical properties.

Conclusions:

• Established optimal data conditions (200 Hz sampling, 170s duration) for nonlinear analysis of PPG signals.
• Confirms the comparable effectiveness of gPPG and rPPG for extracting dynamical health information.
• Enhances the reliability and applicability of PPG-based wearable health monitoring technologies.