Friday, May 6 2022
1:00pm - 3:00pm
Masters Presentation
A comparison of the frequentist and Bayesian spatial scan statistic in disease outbreak monitoring

Early disease outbreak detection enables earlier action for public health measures. Early detection of disease outbreak can prompt early investigation of the causes such as naturally occurring epidemics, bioterrorist attacks, or environmental hazards and early implementation of public health measures. Daniel Neill put forward the Bayesian spatial scan statistic (BSSS) in 2006, claiming that it was a significant improvement over the traditional frequentist method. In this paper, we implement the BSSS method on different outbreak scenarios and evaluate the performances with measures such as detection time and detection rate. We have found that the frequentist scan statistic performs better than the BSSS in common metrics. Possible explanations are discussed. We also propose a hypothesis suggesting that under fair comparison requirements, the frequentist scan statistic cannot be beaten by other methods.
Speaker:Dongdong Lu
Location:Zoom link in email

Download as iCalendar