PPGFlowECG: Latent Rectified Flow with Cross-Modal Encoding for PPG-Guided ECG Generation and Cardiovascular Disease Detection

2025-09-24

AI Summary

Photoplethysmography (PPG) lacks electrophysiological specificity and cannot support definitive cardiovascular diagnosis. To address this, we propose PPGFlowECG—the first two-stage PPG-to-ECG generation framework trained on MCMED, a million-scale clinical paired PPG-ECG dataset. Methodologically, it innovatively integrates CardioAlign, a cross-modal encoder for physiological semantic alignment, and introduces Latent Rectified Flow to model high-dimensional temporal dependencies in the latent space, balancing generation fidelity and interpretability. Experiments demonstrate that the synthesized ECGs faithfully reproduce critical diagnostic features—including waveform morphology, rhythm, and ST-segment characteristics—with high clinical reliability, as validated by cardiologists. Furthermore, PPGFlowECG significantly improves detection accuracy and diagnostic consistency across multiple cardiovascular disease screening tasks. This work establishes a novel paradigm for non-invasive, continuous, and low-cost cardiac health monitoring.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Multi-modal Vision
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

In clinical practice, electrocardiography (ECG) remains the gold standard for cardiac monitoring, providing crucial insights for diagnosing a wide range of cardiovascular diseases (CVDs). However, its reliance on specialized equipment and trained personnel limits feasibility for continuous routine monitoring. Photoplethysmography (PPG) offers accessible, continuous monitoring but lacks definitive electrophysiological information, preventing conclusive diagnosis. Generative models present a promising approach to translate PPG into clinically valuable ECG signals, yet current methods face substantial challenges, including the misalignment of physiological semantics in generative models and the complexity of modeling in high-dimensional signals. To this end, we propose PPGFlowECG, a two-stage framework that aligns PPG and ECG in a shared latent space via the CardioAlign Encoder and employs latent rectified flow to generate ECGs with high fidelity and interpretability. To the best of our knowledge, this is the first study to experiment on MCMED, a newly released clinical-grade dataset comprising over 10 million paired PPG-ECG samples from more than 118,000 emergency department visits with expert-labeled cardiovascular disease annotations. Results demonstrate the effectiveness of our method for PPG-to-ECG translation and cardiovascular disease detection. Moreover, cardiologist-led evaluations confirm that the synthesized ECGs achieve high fidelity and improve diagnostic reliability, underscoring our method's potential for real-world cardiovascular screening.

Problem

• Generating ECG signals from accessible PPG data for clinical use
• Addressing physiological misalignment in cross-modal generative models
• Enabling cardiovascular disease detection through PPG-to-ECG translation

Innovation

• Latent rectified flow generates ECG signals
• Cross-modal encoder aligns PPG and ECG
• Two-stage framework for high-fidelity generation