Biology Reference
In-Depth Information
Limitations of the aberration detection method:
•
An increase in a time series could be due to variability within the
time series such as seasonality and day-of-week effects.
Numerous algorithms can be applied to surveillance data, and, for
•
each algorithm, parameter selection affects the time at which the
algorithm will alert.
Algorithms cannot replace the need for local knowledge and
•
experience.
1.4 Correlation
In terms of correlation, timeliness is defined as the time lag at which the
correlation between two data sources is significant (Johnson et al. 2004).
Specifically, the correlation between two time series is calculated, and then
one of the series is moved in time relative to the other, representing different
time lags. The correlation is calculated for each time lag, and the timeliness
of one data source relative to another is determined by the correlation coef-
ficients that are statistically significant (see Box 1.3).
The cross-correlation function (CCF) calculates a numerical value that
describes the similarity of two curves over a defined period with two identi-
cal curves having a CCF value of one (Suyama et al. 2003). Prior to computing
the CCF, the data have to be normalized to satisfy the assumption of normal-
ity. A significant CCF at a specific time lag “x” indicates that the peak in one
data source occurs “x” time periods before the peak in another data source. A
significant CCF at a lag of zero indicates the peaks occurred at the same time
(Suyama et al. 2003). The Spearman rank correlation is another method used
to determine correlation, which is based on ranking data, and can be applied
when the assumption of normality is violated (Doroshenko et al. 2005).
As shown in Figure 1.3, similar time series have a high correlation (close to
1.0), and dissimilar time series have a low correlation (close to 0). When inter-
preting correlation values, low correlations may be statistically significant
due to the randomness and variability in biosurveillance data sources.
Timeliness has been assessed by correlation methods, including the CCF
(Espino, Hogan, and Wagner 2003; Hogan et al. 2003; Ivanov et al. 2003;
Johnson et al. 2004; Magruder 2003) and the Spearman rank correlation
(Doroshenko et al. 2005).
Espino and colleagues (2003) determined the CCF of regional and state
influenza activity to emergency room telephone triage based on 10 hospitals
in a large city. They found that emergency room telephone triage was 7 days
(correlation 0.25) ahead of state influenza activity and 28 days (correlation
