Biology Reference
In-Depth Information
V. TIME-INDEPENDENT MEASURES: INTERBEAT
INTERVAL DISTRIBUTION, STANDARD DEVIATION,
AND SKEWNESS
As discussed above, the length of sequential RRIs varies from beat to
beat, and the degree of its variability is related to health or illness.
A time-independent approach to investigating the variability of a sample
of RRIs focuses on their statistical distribution (this approach is time-
independent because changing the order of RRIs within the sample will
not change its statistical distribution). In this setting, the length of a RRI
is assumed to be a random quantity with a certain unknown
distribution. A histogram of a certain sample of RRIs depicts one
realization of this distribution. The shape and the symmetry of this
histogram provide information about the properties of the random RRIs.
Changes in the histogram may, therefore, signify changes in health.
We want to develop a measure of how at-risk a baby is for the imminent
onset of sepsis based on HRV. If we examine the panels in Figure 6-3,
some things stand out. In panel A, where the baby is clinically stable,
there is a pattern of ''regular irregularity.'' Compared with panels B and
C, there is a substantial amount of variation throughout the period in the
RRIs, but the variation seems to be under control. In panel B, when the
baby's health has begun to deteriorate, there is very little variation
during most of the period, with a few large decelerations. In panel C, as
health continues to worsen, these changes become more pronounced.
The fact that panels B and C depict RRI data with little variation
suggests that the corresponding histograms will have a relatively low SD.
The isolated decelerations, on the other hand, suggest a certain asymmetry
in the histogram of the RR data. To better understand why, we briefly
review the relation between the following basic descriptors of a
histogram—the mean, the median, and the skewness.
The two most common ways to measure the center of a body of
quantitative data are the mean and the median. The mean is what people
often refer to as the average, and the median is the 50th percentile.
In some ways, the median can be more descriptive of the data's center
because it is not affected by extreme values (as the mean can be). A body
of data is symmetric if the histogram can be divided into two halves that
are mirror images of one another. If the data are symmetric, the mean
and the median coincide. 1 In reality, it would be unusual to find a
completely symmetric body of data, but many phenomena give data that
are approximately symmetric.
Some data that are not symmetric can be described as skewed to the left
or right. To get an intuitive idea of skewness, suppose we begin with
1. It is possible, however, for the mean and the median to coincide when the
data are not symmetric.
Search WWH ::




Custom Search