PENGUIN: General Vital Sign Reconstruction from PPG with Flow Matching State Space Model

TL;DR

PENGUIN is a novel generative framework that reconstructs multiple vital signs from PPG signals with high accuracy, overcoming noise and motion artifacts, and generalizes well across diverse datasets and tasks.

Contribution

We introduce PENGUIN, a flow-matching model that extends state space models for fine-grained, multi-vital sign reconstruction from PPG signals, addressing limitations of existing methods.

Findings

Outperforms task-specific and general-purpose baselines across six datasets.

Successfully reconstructs ECG, respiratory, and ABP signals from PPG.

Demonstrates robustness to noise and motion artifacts.

Abstract

Photoplethysmography (PPG) plays a crucial role in continuous cardiovascular health monitoring as a non-invasive and cost-effective modality. However, PPG signals are susceptible to motion artifacts and noise, making accurate estimation of vital signs such as arterial blood pressure (ABP) challenging. Existing estimation methods are often restricted to a single-task or environment, limiting their generalizability across diverse PPG decoding scenarios. Moreover, recent general-purpose approaches typically rely on predictions over multi-second intervals, discarding the morphological characteristics of vital signs. To address these challenges, we propose PENGUIN, a generative flow-matching framework that extends deep state space models, enabling fine-grained conditioning on PPG for reconstructing multiple vital signs as continuous waveforms. We evaluate PENGUIN using six real-world PPG datasets across three distinct vital sign reconstruction tasks (electrocardiogram reconstruction, respiratory monitoring, and ABP monitoring). Our method consistently outperformed both task-specific and general-purpose baselines, demonstrating PENGUIN as a general framework for robust vital sign reconstruction from PPG.