Estimating respiratory rate in freely moving users using independent component and multi-resolution analysis

Area of Science:

• Biomedical Engineering
• Signal Processing
• Health Monitoring Technologies

Background:

• Non-invasive health monitoring is crucial for continuous, comfortable vital sign tracking.
• Contactless radar vital sign sensing faces challenges with noisy signals and subject movement.
• Accurate extraction of physiological data from radar signals under free-movement conditions remains a significant hurdle.

Purpose of the Study:

• To introduce and validate a novel framework for estimating respiratory rate (RR) from radar signals.
• To address the challenges of accurate vital sign extraction in freely moving subjects.
• To improve the accuracy of respiratory rate estimation in real-world, dynamic environments.

Main Methods:

• A novel framework combining Independent Component Analysis (ICA) and Empirical Wavelet Transform (EWT) was developed.
• ICA was used to automatically select physiologically relevant components from radar signals.
• EWT served as an adaptive Multi-Resolution Analysis (MRA) for signal decomposition and reconstruction to enhance peak detection.

Main Results:

• The proposed ICA-EWT framework demonstrated effectiveness in estimating respiratory rate (RR) under free-movement conditions.
• Pearson's correlation coefficient revealed a strong correlation (r = 0.94) between estimated and reference RR.
• Bland-Altman analysis indicated a low mean error of -0.41 breaths per minute.

Conclusions:

• The ICA-EWT framework shows significant promise for accurate, non-invasive respiratory rate estimation in real-world settings.
• The method is effective even when subjects are moving freely, overcoming a major limitation of current radar-based systems.
• Future improvements may involve optimizing radar placement and mitigating signal interference to further enhance accuracy.