The West Nile virus is a mosquito borne agent that has infected hundreds of thousands of birds and mammals and thousands of humans since its arrival in the United States in 1999.
Since its introduction to the Americas, the disease has received renewed attention by researchers. A significant part of this research, dealt with the identification of proxies for WNV presence in specific areas so that public health officials could intervene with control measures and reduce the risk of infection in humans. One of the proxies used is the death of birds in close spatial and temporal proximity.
The Dynamic Continuous Area Space-Time (DYCAST) system, which identifies local WNV risk areas, is based on the hypothesis that WNV propagation occurs in a cycle between birds (hosts) and mosquitoes (vectors). The cycle eventually amplifies resulting in an increased pool of the virus among birds. Mosquitoes that feed on both birds and humans are then more likely to carry and transmit the virus to humans.
DYCAST models the amplification process as a continuum in space and time. The quantitative analysis which models this transmission involves the design and implementation of a system that is based on a geographic model that uses a Knox test to capture the nonrandom space-time interaction of dead birds, as an indicator of an intense West Nile amplification cycle, within a 1.5 mile (2.41 km) buffer area and 21-day window. The Knox analysis is implemented as an interpolation function to create a surface of probabilities over a grid overlaying the area. The system is implemented in a geographic system and used operationally in New York City and the City of Chicago where was modified with an unconditional extension of the Knox method to address weaknesses inherent in the Knox methodology and significance testing. The results from the traditional Knox method and the unconditional extension of the Knox method were evaluated using a novel space-time implementation of the Kappa Index of agreement for the Chicago data. The Kappa results were used to quantify the success of each method and offer evidence for the role of dead birds in human West Nile virus transmission. It is shown that the discriminatory power of the unconditionally extended Knox is greater and that the two tests produce significantly different results.