Abstract
We have developed an automated system that reads emergency department clinical reports and constructs models of multiple, possibly overlapping outbreaks. The system relies on a Bayesian scoring metric and search algorithms to find appropriate models. The system has been tested on simulated and actual data with good results.