Physiologically Interpretable Photoplethysmography Features Associated With Peripheral Artery Disease

Figures

• 

  Figure 1.. Falling-edge timing features are associated with ankle-brachial index (ABI) in a segment-specific manner (5,282 signals, N=2,599). Top row: each feature (red) defined on a representative normalized toe photoplethysmography (PPG) pulse. Bottom row: each feature plotted against ABI per leg for the combined dataset, with linear fit and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}$\end{document}. The red marker and arrow in (a) locate the example pulse’s feature value within the population. (a) Time to 50% amplitude on the falling edge, measured as absolute time from pulse onset (feature A), is negatively associated with ABI (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}=0.299$\end{document}, p< 0.0001). (b) Early-decay 90-60% segment duration (feature B) is negatively associated (\documentclass[12pt]{minimal}…

• 

  Figure 2.. Localized slope and integrated area features reflect a common temporal shift (5,282 signals, N=2,599). Top row: each feature (red) defined on a representative normalized toe photoplethysmography (PPG) pulse. Bottom row: each feature plotted against ankle-brachial index (ABI) per leg for the combined dataset, with linear fit and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}$\end{document}. (a) Slope at 37% normalized pulse duration (feature E) is negatively associated with ABI (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}=0.256$\end{document}, p< 0.0001) and (b) slope at 68% (feature F) is positively associated (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs}…

• 

  Figure 3.. Average toe photoplethysmography (PPG) waveform morphology across ankle-brachial index (ABI) severity strata (pooled cohorts from combined dataset of 5,282 signals, N=2,599). (a) Mean normalized waveform for each ABI stratum, showing a graded progression toward a later peak and slower, more prolonged decay with decreasing ABI. (b) Mean waveforms for severe peripheral artery disease (PAD) (ABI ≤ 0.5) and normal (1.0 < ABI ≤ 1.4) with ±1 standard deviation (SD) bands, showing the contrast persists relative to within-group variability. n per stratum as labeled.

• 

  Figure 4.. Amplitude-based photoplethysmography (PPG) features show no meaningful association with ankle-brachial index (ABI) (La Jolla cohort, 3,457 signals, N=2,226). (a) Raw baseline-to-peak amplitude versus ABI (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}=0.047$\end{document}). (b) Within-patient inter-leg amplitude ratio versus inter-leg ABI difference (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}=0.031$\end{document}). (c) Within-patient inter-leg amplitude ratio versus ABI (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} $R^{2}=0.012$\end{document}). Dashed lines show linear fits.