Abstract
Biosurveillance involves monitoring measures of diagnostic and pre-diagnostic ac-tivity for early detection of disease outbreaks. Modern biosurveillance data include daily countsof diagnostic evidence such as lab results, and pre-diagnostic health seeking behavior suchas medication sales. A serious challenge to research in the field of biosurveillance is the lackof available authentic data to researchers. This significantly limits the possibility of algorithmdevelopment and evaluation and hinders the comparison of methods across different groupsof researchers. Since biosurveillance datasets are usually proprietary and tightly held by theirowners, an alternative is generating simulated or semi-authentic data that are similar to au-thentic datasets. This paper describes a method for simulating multivariate biosurveillancetime series, in the form of daily counts from multiple biosurveillance series, by using statisticsfrom authentic biosurveillance data. Moreover, it uses statistical methods to test the validity ofthese simulated series, testing whether they could reasonably have come from the same dis-tribution as the authentic series. We make simulator software and datasets publicly available.