Although there is no substitute for actual data, there are times when simulated data can be very useful, particularly for evaluating how systems and methods will perform under conditions other than that which has already been observed. In addition, simulation gives researchers and practitioners control that is simply not achievable with real data. For example, in real data, it is very difficult – often impossible – to determine whether a particular data stream contains one or more outbreaks. And even when the existence of an outbreak is known, it is generally impossible to definitively determine when it started and ended.

Thus, although some reject simulated data under the assumption that they cannot mimic all the features of real biosurveillance data, the use of simulation – which must be based on idealizations of the real world to some degree – is useful precisely because:

1. • it permits definitive performance evaluations and comparisons under known conditions, and

2. • it can often allow one to assess how the various data features affect performance.