Year: 2025
Type: article
Abstract: Respiratory rate (RR) is an important vital sign indicating various pathological conditions, such as clinical deterioration, pneumonia, and adverse cardiac arrest. Traditional RR measurement methods are normally intrusive and inconvenient for ubiquitous continuous monitoring. There have been studies on RR estimation by extracting respiratory modulated components (RMCs) from wearable accessible noninvasive cardiovascular signals, such as electrocardiogram (ECG) or/and photoplethysmogram (PPG), with RR estimated from each RMC or fused RMCs derived from either ECG or PPG. However, there is few study on robust continuous RR estimation with the combination of all kinds of RMCs from both ECG and PPG in the time domain. In this study, we propose the temporal fusion of RMCs extracted from both ECG and PPG signals to estimate RR with the aim to improve estimation performance. We extracted six RMCs from ECG and PPG, identified those RMCs of high quality with the respiratory quality index, fused the identified ones into one respiratory signal with principal component analysis, and estimated the RR from the fused signal. Validation on two public datasets - the Capnobase dataset (42 subjects) and the BIDMC dataset (53 subjects) - showed that the proposed method attained a mean absolute error (MAE) of 1.39 breaths/min and 3.29 breaths/min for RR estimation, respectively, achieving an average 11.61% reduction in MAE compared to existing state-of-the-art approaches. This demonstrates that temporal fusion of the RMCs of wearable ECG and PPG can improve the performance of RR estimation.
Source: PLoS ONE
Authors: Yuxuan Lin, Xinyue Song, Yan Zhao, Chunlin Zhang, Xiaorong Ding
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,382
Awards: G5075172614, G7200226891
OpenAlex ID: W4411358995