A Resource-Efficient Time-Domain-Based Algorithm to Estimate Respiration Rate From Single-Lead ECG Signal
This study introduces a novel, computationally efficient time-domain (TD) algorithm for accurate breath rate (BR) estimation from single-lead ECG signals, designed for wearable devices. The proposed algorithm uses statistical TD parameters—mean, prominence, and distance (MPD)—t...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Open Journal of Instrumentation and Measurement |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10915585/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study introduces a novel, computationally efficient time-domain (TD) algorithm for accurate breath rate (BR) estimation from single-lead ECG signals, designed for wearable devices. The proposed algorithm uses statistical TD parameters—mean, prominence, and distance (MPD)—to detect valid respiratory peaks in ECG-derived respiration (EDR) signals. The performance of the MPD algorithm was evaluated using two datasets: 1) a benchmark database containing ECG acquired during dynamic activities and 2) a real-time dataset comprising ECG signals from five subjects performing dynamic activities, including standing, jogging, and recovery. Comparative analysis against state-of-the-art TD methods, such as count-orig, zero-crossing detection, peak detection, and adaptive threshold techniques, demonstrates the superiority of MPD in both accuracy and computational efficiency. On the benchmark dataset, MPD achieved a mean absolute error (MAE) of 3.66 bpm and mean absolute percentage error (MAPE) of 23.69%, outperforming the Count-Orig method (MAE = 5.09 bpm, MAPE = 32.76%). For real-time data, MPD further demonstrated robust performance with an MAE of 1.53 bpm and MAPE of 7.25%. The algorithm’s design simplicity, combined with its ability to handle spurious peaks and varying signal conditions, makes it particularly suitable for resource-constrained wearable applications. Its high accuracy, low computational demands, and adaptability across activity conditions underscore its potential for continuous, real-time respiratory monitoring in diverse scenarios. |
|---|---|
| ISSN: | 2768-7236 |