Validation of time-domain heart rate variability based on data from Actiheart recordings
# Background
From the Actiheart recordings in ADDITION-PRO study, there are heart rate data every 30 seconds with a 95 % prediction interval. We have developed an algorithm that calculates HRV, which calculates the distribution of interbeat interval (IBI) in each 30-second epoch and includes the generated interval in the r package called RHRV. The document validates the algorithm on data set from physionet with all available IBI.
# Use case
As many wearable devices measure periodic IBI or mean heart rate with a prediction interval, we need to test whether we can derive standard time-domain HRV measures from incomplete IBI during the whole worn period. Frequency-domain HRV measures are not relevant for this purpose, as we would require access to all IBI data during the period.
The algorithm for this is based on the RHRV package (version 4.2.7) and creates an automatic pipeline to calculate time-domain HRV based on the input of sequences of IBI. Contact Jonas R Schaarup (jfrscha@ph.au.dk), for further description of the algorithm or if needed help with implementation. With full access to IBI, frequency-domain measures can be implemented in this pipeline.
Reference to RHRV package:
*Heart Rate Variability Analysis with the R package RHRV, C.A. García, A. Otero, X.A. Vila, M.J. Lado, L. Rodríguez-Liñares, J.M. Presedo, and A.J. Médez. Springer International Publishing, 2017. Springer book page*
# Aim
To validate time-domain HRV measures based on 30 second epoch including mean heart rate data and 95 % prediction interval with time-domain HRV measures with full access to all IBI
# Data description
We include 76 healthy subject with 24-hour holter recordings. Description of the full data set can be found in the reference below.
Reference to dataset:
*Irurzun, I. M., Garavaglia, L., Defeo, M. M., & Thomas Mailland, J. (2021). RR interval time series from healthy subjects (version 1.0.0). PhysioNet. https://doi.org/10.13026/51yd-d219.*
# Methods
We calculated 30 second mean HR with a 95% prediction interval based on the last 8 heart beats in the 30 second segment.
We tested the algorithm for HRV measures based generated IBI from the 30 sec mean HR and compared it HRV calculated from the actual heart beats interbeat intervals. We plotted the generated HRV and the actual HRV and performed spearman correlation between the two HRV measures.
