Compressed ECG sensing based heart rate asymmetry analysis for energy-constrained fast health monitoring

Abstract

The Poincaré plot based non-linear heart rate variability (NLHRV) parameters are helpful to understand the heart rate asymmetry (HRA). These are also identified as HRA descriptors. The HRA represents a physiological phenomenon that signifies the persistent difference in how heart rate decelerations and heart rate accelerations affect the functioning of the human cardiovascular system. Most of the HRA descriptors are variance-based descriptors. These HRA descriptors can be evaluated using the time-series signal consisting of inter-beat intervals (IBI). Rather than deriving the IBI signal using high-resolution R-peak detection for an electrocardiogram (ECG) signal, compressed sensing (CS) based R-peak detection for IBI signal computation is more suitable especially when short-term fast cardiac heath monitoring is our prime concern. For this comparative study, 48 ECG signal records of MIT-BIH arrhythmia (MIT-BIH-A) database and 18 ECG signal records of MIT-BIH normal sinus rhythm (MIT-BIH-NSR) database are used. A set of 23 HRA descriptors has been evaluated using both R-peak detection methods for 1 min duration signal from above mentioned databases. For comparative HRA analysis, we have observed more than 95% similar values for all 23 HRA descriptors for both databases. Comparative error analysis is presented in this paper to validate the suitability of the compressed ECG sensing based NLHRV analysis with respect to the conventional high resolution ECG based non-linear HRV analysis. Role of short-term fast ECG sensing using wearable IoT devices is significant and short-term NLHRV analysis is helpful for the fast health monitoring. © 2024 Elsevier Ltd