Skip directly to search Skip directly to A to Z list Skip directly to page options Skip directly to site content

Volume 15, Number 4—April 2009


Enhancing Time-Series Detection Algorithms for Automated Biosurveillance

Jerome I. TokarsComments to Author , Howard Burkom, Jian Xing, Roseanne English, Steven Bloom, Kenneth Cox, and Julie A. Pavlin
Author affiliations: Centers for Disease Control and Prevention, Atlanta, Georgia, USA (J.I. Tokars, J. Xing, R. English); The Johns Hopkins University, Baltimore, Maryland, USA (H. Burkom); Science Applications Incorporated, San Diego, California, USA (S. Bloom); Department of Defense, Washington, DC, USA (K. Cox); Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand (J.A. Pavlin)

Main Article

Table 2

Mean absolute residual, by method and dataset, for selected BioSense data used in algorithm modification study*

Stratification of baseline by weekday vs. weekend Mean absolute residual
Department of Defense
Hospital emergency department
Count Rate Count Rate
Unstratified 4.2 2.4 2.2 2.0
Stratified 2.4 2.2 2.3 2.0

*The count method uses only numerator data; the rate method uses numerator and denominator data. Because varying the baseline duration did not affect residuals (data not shown), all calculations shown here use a baseline duration of 7 days.

Main Article