Trend Strips: A new tool to analyze RR time series

There are many different tools and parameter that can be used to analyze RR time series, as the Correlation Dimension, the Lyapunov Exponents, the Higher Reconstruction Step. Here we address the problem of discriminating between two populations: one with some cardiac disease and another with healthy people, in a no invasive way, through RR time series. The mathematical tool used to analyze the series is called Trend Strips (TS) and was proposed recently in the scientific literature. TS are a sequence of “0” and “1”, where “1” means that the value in some position is greater than the value in a previous position in the series, while “0” means the opposite. We can use TS of any size, where the size is the amount of “0” and “1” considered. If we choose TS of size 4 we will have 16 different sequences at our disposal, from “1111” to “0000”. What we do, then, is to compute the relative frequency of occurrence of each of them in a particular time series and to compare the frequencies in both groups. We collected RR time series from two groups of men at rest: one group with a cardiac chagasic disease (24 individuals) and a second one of healthy people (21 individuals). We analyzed TS of sizes from 2 to 5 and found significant TS that discriminate the two groups at all sizes. The statistical test used was a t-test and some of the TS useful, with a p < 0.01 were: (a) for n=2, TS “00” (p=0.0009); (b) n=3, TS “101” (p=0.0055) and TS “000” (p=0.0061); (c) for n=4, TS “1011” (p=0.0059), TS “1000” (p=0.0071) and TS “0001” (p=0.0073), etc. The set of significant TS can be useful, than, for diagnostic purposes.