Physiologically Interpretable Photoplethysmography Features Associated With Peripheral Artery Disease

Abstract

Objective: Peripheral artery disease (PAD) is a common and frequently underdiagnosed cardiovascular condition. Photoplethysmography (PPG) is a widely available optical technique with strong potential to support peripheral vascular assessment, yet its feature-level relationship to PAD severity has not been systematically characterized at scale. This study aims to characterize physiologically informed, feature-level relationships between PPG pulse morphology and PAD severity using a large clinical dataset benchmarked against ankle-brachial index (ABI). Results: Using a large, multi-site clinical dataset consisting of 5,282 toe PPG waveforms from 2,599 unique patients, we systematically evaluated a broad set of PPG features. Features capturing waveform timing and shape exhibited consistent associations with disease severity, including pulse broadening and delayed waveform progression. In contrast, raw signal amplitude did not reliably associate with ABI. Feature associations were consistent across complementary feature classes, and exclusions did not introduce demographic bias. Conclusions: These findings show that PAD severity is reflected in coordinated changes in PPG pulse timing and shape, providing physiological context for waveform features. This work provides a feature-level reference framework for physiologically interpretable use of PPG in peripheral vascular assessment.