Photoplethysmogram beat detection using Symmetric Projection Attractor Reconstruction.

Area of Science:

• Physiology
• Dynamical Systems Theory
• Signal Processing

Background:

• Photoplethysmogram (PPG) signals are widely used for physiological monitoring.
• Accurate beat detection in PPG is crucial for reliable health assessments.
• Existing beat detection algorithms face challenges with signal variability and noise.

Purpose of the Study:

• To introduce a novel beat detection method for PPG signals.
• To evaluate the performance of the proposed method across different physiological states.
• To compare the novel method against existing beat detection algorithms.

Main Methods:

• Generation of a two-dimensional phase space attractor from PPG signals using Symmetric Projection Attractor Reconstruction (SPAR).
• Definition of a Poincaré section through the origin of the phase space for beat detection.
• Optimization of the Poincaré section to account for baseline drift.
• Validation using the WESAD dataset encompassing baseline, stress, amusement, and meditation conditions.

Main Results:

• Median F1 scores achieved: 74.3% (Baseline), 63.0% (Stress), 93.6% (Amusement), 97.7% (Meditation).
• The SPAR method demonstrated superior performance compared to an earlier version.
• The method's performance was comparable to top-tier algorithms in a benchmark study.
• Improved accuracy in inter-beat interval measurements for resting subjects compared to two other methods.

Conclusions:

• The SPAR method offers a robust and accurate approach for beat detection in PPG signals.
• The method shows significant potential for application in diverse physiological monitoring scenarios.
• SPAR provides a competitive alternative to existing beat detection algorithms, particularly in challenging conditions.