Biology Reference
In-Depth Information
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
9
8
7
6
5
4
3
2
1
0
Days prior to sepsis
FIGURE 6-12.
Changes in HR sample entropy before sepsis.
X. COMBINING VARIOUS MEASURES OF HEART RATE
VARIABILITY ABNORMALITY
Abnormalities in HRV characteristic of illness can be identified by
comparing the HRV parameters we have discussed (SD, skewness, SA,
and SampEn) using more advanced mathematical models, such as
logistic regression models, neural networks, multiple variable analysis,
nearest neighbor analysis, or other predictive mathematical instruments.
It is most important to track the data of each at-risk infant and compare
sequential HR time series. Experiments show that changes observed
in sequential 4096-beat time series tracked over time provide a good
indication and reliable warning of upcoming episodes of sepsis. In order
to conduct such experiments, the HR of each at-risk baby is recorded
continuously, using bedside monitors.
A large clinical trial involving 316 neonates in the University of Virginia
NICU and 317 neonates in the Wake Forest University NICU tested and
validated the utility of the HRV characteristics presented in this chapter.
In this trial, clinical data were prospectively collected, and RRIs were
continuously recorded in all infants in these two NICUs who stayed for
more than 7 days. Episodes of sepsis and sepsis-like illness were defined
as acute clinical deteriorations that prompted physicians to obtain blood
cultures and start antibiotics. During the trial, 273 such episodes were
observed in 194 of the infants, a staggering number, which demonstrates
the high risk for sepsis in these premature babies.
A predictive statistical model was developed using data from the
derivation cohort in the University of Virginia NICU, and then tested on
the validation cohort at Wake Forest University. An HR characteristics
(HRC) index was defined as the output of a regression model combining
the RRI characteristics described in this chapter to predict sepsis and
