Estimation of the uncertainty in time domain indices of RR time series

Abstract

A method for estimating the uncertainty in time-domain indices of RR time series is described. The method relies on the central limit theorem that states that the distribution of a sample average of independent samples has an uncertainty that asymptotically approaches to the sample standard deviation divided by the square root of the number of samples. Because RR time series cannot be characterized by a set of independent samples, we propose to estimate the uncertainty of indices by computing them in blocks that satisfy that the obtained partial indices are independent. We propose a methodology to search sets of independent partial indices and apply this methodology to the estimation of the uncertainty in the mean RR, SDRR, and r-msDD indices. The results show that the uncertainty can be higher than the 10% of the index for the SDRR and even higher for the r-msDD. Moreover, a statistical test for the difference of two indices is proposed.