Biology Reference
In-Depth Information
VIII. VALIDATING THE PROPERTIES OF SAMPLE
ASYMMETRY THROUGH COMPUTER SIMULATION
In addition to testing with real data, a commonly accepted strategy for
validating a new statistical measure is to simulate data with certain
properties and then investigate whether the new measure depicts these
properties accurately. In the case of SA, we need to confirm its ability
to distinguish, in a graded fashion, HR time series with varying degrees
of transient decelerations. So, we constructed simulations consisting
of a baseline signal with the same frequency components as clinically
observed neonatal HR data, and then added to this signal scaled
versions of clinically observed deceleration. In this way, we were able
to simulate a wide range of HRV abnormalities. Figure 6-9 shows an
example of simulated data with baseline SD 7 milliseconds and four
decelerations of magnitude 100 milliseconds, corresponding to
moderately reduced HRV with subclinical decelerations from 150 beats
per minute (RRI 400 msec) to 120 beats per minute (RRI of 500 msec).
This is a realistic simulation—inspection of Figure 6-3(B) suggests
a deceleration might easily be 20 times larger than the SD of a stable,
low-variability neonatal HR.
The results are plotted in Figure 6-10. SA is plotted as a function of the
height of the deceleration measured in SD of the baseline signal.
We simulated an increasing number of decelerations, from 0 to 8.
Figure 6-10 shows that both larger and more frequent decelerations lead
to increasing values of SA, with values exceeding 10 in many
conditions. For the example in Figure 6-9, with four decelerations
500
450
400
350
0
1000
2000
3000
4000
Beat number
FIGURE 6-9.
Simulated heart rate data with baseline variability with SD
7 milliseconds and 4 decelerations.
(From Kovatchev, B. P., Farhy, L. S., Hanging, C., Griffin, M. P., Lake, D. E., Moorman, J. R. [2003].
Sample asymmetry analysis of heart rate characteristics with application to neonatal sepsis and
systemic inflammatory response syndrome. Pediatric Research, 54, 892-898. Used by permission.)
¼
