pyPPG

pyPPG enables long-term continuous photoplethysmography (PPG) time-series analysis through beat detection, fiducial point identification, and digital biomarker computation for assessing vascular dynamics and physiological parameters.

Key Features:

• Beat Detection: An improved PPG peak detector achieved an F1-score of 88.19% on 2054 adult polysomnography recordings totaling over 91 million reference beats.
• Fiducial Point Detection: The detector identifies multiple fiducial points with a mean absolute error of less than 10 ms, validated using more than 3000 manually annotated fiducial points by two annotators.
• Biomarker Engineering: The toolbox computes 74 PPG biomarkers derived from detected fiducial points for standardization and validation of digital PPG biomarkers.

Scientific Applications:

• Vascular Dynamics Assessment: PPG signal analysis for evaluating blood volume changes in tissues and vascular function.
• Disease Etiology Research: Studying PPG time-series variability to enhance understanding of disease manifestations and etiology.
• Data-Driven Model Training: Biomarker engineering for training data-driven models in research and clinical applications.

Methodology:

The toolbox uses a standard finger-based transmission pulse oximeter for signal acquisition, implements an improved PPG peak detection algorithm, and detects fiducial points for biomarker computation.