A collaboration of veterinary epidemiologists, laboratorians, and statistical analysts has designed a laboratory-based health surveillance system for horses in Colorado. Initial efforts focused on 12 years of data from three state labs. Exploratory analysis, stakeholder input, and discovery of laboratory workflow details were applied to refine syndrome groups and filter test order records to eliminate alerting bias. To tailor alerting methods for each syndrome, we constructed an algorithm testbed and a stochastic injection process based on target signal types chosen by the epidemiologists based on known disease progression. Testbed results yielded recommended methods, settings and thresholds for each syndrome.