Bayesian group testing regression models for spatial data

Rongjie Huang,Alexander C McLain,Brian H Herrin,Melissa Nolan,Bo Cai,Stella Self

DOI: https://doi.org/10.1016/j.sste.2024.100677

Abstract:Spatial patterns are common in infectious disease epidemiology. Disease mapping is essential to infectious disease surveillance. Under a group testing protocol, biomaterial from multiple individuals is physically combined into a pooled specimen, which is then tested for infection. If the pool tests negative, all contributing individuals are generally assumed to be uninfected. If the pool tests positive, the individuals are usually retested to determine who is infected. When the prevalence of infection is low, group testing provides significant cost savings over traditional individual testing by reducing the number of tests required. However, the lack of statistical methods capable of producing maps from group testing data has limited the use of group testing in disease mapping. We develop a Bayesian methodology that can simultaneously map disease prevalence using group testing data and identify risk factors for infection. We illustrate its real-world utility using two datasets from vector-borne disease surveillance.