Evaluation of HRV from Repeated Measurements of PPG and Arterial Blood Pressure Signals

Abstract

Heart rate variability (HRV) is a valuable noninvasive indicator of the autonomic nervous system. Guidelines prescribe that HRV is calculated from RR intervals of the ECG signal. With the development of wearables and the availability of unobtrusive photoplethysmography (PPG) sensors inter-beat intervals (IBI) are considered as a viable HRV source. We compare short-term HRV from ECG, PPG and arterial blood pressure (ABP), which is a pulse signal of the best obtainable quality. We evaluate several characteristic points to derive IBI and validate the accuracy of individual HRV metrics for each signal.

We analysed 69 extracts (5 min) of rest data from 25 healthy volunteers recorded three times within two hours. ABP is superior to PPG, but both can be used for pulse rate variability (PRV) of subjects at rest, though not all metrics are equally reliable. The most error-prone features are high frequency power and pnn50 with relative errors of 22% (16%) and 30% (12%) for PPG (ABP). Bland-Altman plots reveal steeper regression trend for PPG-based PRV and tendency to overestimate low RMSSD and SDNN values more. Intra-subject volatility of explored metrics exceeds the errors by at least several times: absolute power features may variate by more than 100%, but their mean errors are up to 20%.