Directional spatial individual level models of infectious disease transmission

Individual level models (ILMs) are a class of mechanistic models that are widely used to infer spatial infectious disease transmission dynamics. These models can incorporate individual level covariate information, accounting for population heterogeneity, to model the spread of an infectious disease over time and space. In many epidemic systems, transmission risk can be directionally-dependent, as well as distance-dependent. This might be due to migration and behaviour patterns in human and animal epidemics, or due to wind and water flow patterns in plant epidemics. Spatial ILMs have not traditionally considered these directional tendencies. We propose a class of ILMs that take both spatial and directional dynamics of disease transmission into account to provide a better fit to epidemic data. We evaluate the performance of these directional ILMs in a simulation study. These methods are also applied to two real epidemic datasets: one from a tomato spotted wilt virus greenhouse experiment; and another from the 2001 foot and mouth disease livestock epidemic in the UK. It is shown that directional ILMs can provide a better fit to epidemic data than purely spatial ILMs in many scenarios.