A Spatiotemporal Analysis of Syndromic Data for Biosurveillance

6 Conclusion We have shown in this chapter not only how to look at the number of patients entering into the surveillance system, but also to consider from whence they came. The statistic used to measure the goodness-of-fit of the spatial distribution we consider is the M-statistic. This statistic has the advantage that its distribution, for large numbers of patients, is approximately independent of the number of patients, so it affords us a two-dimensional statistic (one spatial and one temporal) to simultaneously evaluate departures from normalcy. The M-statistic is based on the distances between patients. We use the fact that in practice this distance distribution is stationary and deviations from this stationarity would indicate a disturbance that might prove worth investigating. As research into distance-based methods and other methods of spatiotemporal methods continues, visualization and disease mapping will become increasingly important since it provides public health officials and decisionmakers additional information, possibly on a real-time or near real-time basis. Our approach to disease mapping, a logical outgrowth of our work on distance-based methods, can be seen as one approach to this important problem.