On the Accuracy of Heart Rate Variability Measures from Undersampled RR Interval Time Series

The heart rate variability (HRV) is the variation in the pulsing frequency of the human heart. Measuring this parameter can reveal important information on the real-time interaction between the autonomic nervous system and the cardiocirculatory system. It can provide useful insight on an individual's state of stress or well being even outside a clinical setting, thanks to inexpensive and unobtrusive wireless sensors and the computing capability of our smartphones. These sensors are battery operated, so their current sampling frequency may be inadequate to measure HRV parameters. In this article, we focus on the accuracy of the detection of the RR interval time series (the instantaneous heart rate), and we investigate the sampling frequency needed to obtain a given level of accuracy for different signals and HRV parameters. We provide the guidelines for an opportunistic choice of the sampling rate to achieve a desired accuracy, and we propose a series of techniques to improve this accuracy with a fixed sampling frequency. We show the effectiveness of the proposed techniques in terms of relative error reduction and energy savings for the wireless sensors.