The purpose of a biosurveillance system is not limited to case and outbreak detection. It should also support outbreak investigation and characterization. Many of the algorithms discussed in Part II can provide information about key outbreak characteristics, including the following:
• Set of affected individuals
• Geographic scope
• Causative organism or toxin
• Incubation period
• Source if other than a person
• Route of transmission
• Host characteristics that explain why some individuals are more susceptible
• Relevant environmental factors
Each characteristic is an uncertain quantity (at least early during an outbreak) that algorithms can potentially infer from biosurveillance data. The algorithms that perform this inference can be evaluated using the same techniques described above for measuring sensitivity, specificity, and timeliness. To our knowledge, no such evaluations have been conducted. Therefore, many questions remain open for future research to determine what kind of data sets could be used to gain insight into the performance of such algorithms, with what accuracy can they infer, automatically, each quantity for different types of outbreaks, and what additional surveillance data or other types of data would improve the accuracy of their inferences.
Was this article helpful?