Public health agencies in the United States have implemented hundreds of syndromic surveillance systems at a cost of hundreds of millions of dollars. Despite the accelerating enthusiasm for this approach, however, there are remarkably few published evaluations of outbreak detection through syndromic surveillance. To meet this need and building on our earlier work to create a scalable architecture for configuring biosurveillance methods,2 we are developing a computational test bed that can draw on real-world data sources and that will allow users to configure, run, and evaluate alternative surveillance methods. The test bed will lower the barriers to evaluation tremendously.
