Machine learning to refine decision making within a syndromic surveillance service.

Within syndromic surveillance systems, machine learning techniques have the potential to make risk assessment following statistical alarms more automated, robust, and rigorous. However, our results also highlight the importance of specialist human input to the process.