INN-PAR: Invertible Neural Network for PPG to ABP Reconstruction

2024-09-13

IEEE International Conference on Acoustics, Speech, and Signal Processing

AI Summary

Existing deep learning–based PPG-to-ABP reconstruction methods suffer from waveform distortion and elevated blood pressure estimation errors due to information loss during the irreversible mapping process. To address this, we propose a gradient-aware invertible neural network (INN) framework that jointly models both raw PPG/ABP signals and their first-order derivatives—a novel formulation in this domain. We further introduce a multi-scale convolutional module (MSCM) to explicitly enhance high-frequency component reconstruction. Evaluated on two public benchmark datasets, our end-to-end approach achieves state-of-the-art performance: ABP waveform similarity distance (WSD) improves by 12.6%, while mean absolute errors for systolic and diastolic blood pressure drop to ±2.1 mmHg and ±1.7 mmHg, respectively. This work represents the first successful application of gradient-guided invertible modeling to noninvasive, continuous blood pressure reconstruction.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Hardware-aware ML
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Non-invasive and continuous blood pressure (BP) monitoring is essential for the early prevention of many cardiovascular diseases. Estimating arterial blood pressure (ABP) from photoplethysmography (PPG) has emerged as a promising solution. However, existing deep learning approaches for PPG-to-ABP reconstruction (PAR) encounter certain information loss, impacting the precision of the reconstructed signal. To overcome this limitation, we introduce an invertible neural network for PPG to ABP reconstruction (INN-PAR), which employs a series of invertible blocks to jointly learn the mapping between PPG and its gradient with the ABP signal and its gradient. INN-PAR efficiently captures both forward and inverse mappings simultaneously, thereby preventing information loss. By integrating signal gradients into the learning process, INN-PAR enhances the network's ability to capture essential high-frequency details, leading to more accurate signal reconstruction. Moreover, we propose a multi-scale convolution module (MSCM) within the invertible block, enabling the model to learn features across multiple scales effectively. We have experimented on two benchmark datasets, which show that INN-PAR significantly outperforms the state-of-the-art methods in both waveform reconstruction and BP measurement accuracy. Codes can be found at: https://github.com/soumitra1992/INNPAR-PPG2ABP.

Problem

• Non-invasive continuous blood pressure monitoring
• Preventing information loss in PPG-to-ABP reconstruction
• Enhancing accuracy with invertible neural networks

Innovation

• Invertible neural network for PPG-to-ABP reconstruction
• Multi-scale convolution module enhances feature learning
• Signal gradients integrated to prevent information loss