MobiVital: Self-supervised Time-series Quality Estimation for Contactless Respiration Monitoring Using UWB Radar

2025-03-14

AI Summary

In ultra-wideband (UWB) radar-based wireless respiration monitoring, extracted respiratory waveforms frequently suffer from distortion and polarity inversion, severely limiting their utility in clinical applications—such as disease diagnosis and rehabilitation training—that demand high-fidelity waveform reconstruction. To address this, we propose the first self-supervised autoregressive waveform extraction framework specifically designed for UWB radar, integrating a biophysically inspired algorithm for automatic polarity inversion detection and correction to achieve end-to-end waveform fidelity enhancement. Furthermore, we publicly release the first synchronized multimodal dataset comprising 12 subjects × 24 hours of concurrent UWB radar and wearable ground-truth respiratory signals, enabling precise temporal alignment and standardized benchmarking. Experiments demonstrate that our method improves respiratory waveform fidelity by 7–34% over state-of-the-art baselines, yielding significant gains in downstream tasks—including respiratory rate estimation.

Technology Category

Intelligent Robots: Multimodal Perception & Sensor FusionMachine Learning: Multimodal LearningComputer Vision: Multi-modal Vision

Application Category

Responsible Web: Machine-in-the-loop, human agency and autonomySearch and Retrieval-Augmented AI: Web evaluation methodologies and metricsUser Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Respiration waveforms are increasingly recognized as important biomarkers, offering insights beyond simple respiration rates, such as detecting breathing irregularities for disease diagnosis or monitoring breath patterns to guide rehabilitation training. Previous works in wireless respiration monitoring have primarily focused on estimating respiration rate, where the breath waveforms are often generated as a by-product. As a result, issues such as waveform deformation and inversion have largely been overlooked, reducing the signal's utility for applications requiring breathing waveforms. To address this problem, we present a novel approach, MobiVital, that improves the quality of respiration waveforms obtained from ultra-wideband (UWB) radar data. MobiVital combines a self-supervised autoregressive model for breathing waveform extraction with a biology-informed algorithm to detect and correct waveform inversions. To encourage reproducible research efforts for developing wireless vital signal monitoring systems, we also release a 12-person, 24-hour UWB radar vital signal dataset, with time-synchronized ground truth obtained from wearable sensors. Our results show that the respiration waveforms produced by our system exhibit a 7-34% increase in fidelity to the ground truth compared to the baselines and can benefit downstream tasks such as respiration rate estimation.

Problem

• Improves quality of respiration waveforms from UWB radar data
• Detects and corrects waveform inversions using biology-informed algorithm
• Enhances fidelity of respiration waveforms for better disease diagnosis

Innovation

• Self-supervised autoregressive model for waveform extraction
• Biology-informed algorithm to correct waveform inversions
• Release of 12-person, 24-hour UWB radar dataset