Abstract

For directly transmitted infectious diseases such as influenza, understanding how best to incorporate human mobility into spatial transmission models is needed for better epidemic prediction and control. The gravity model is a popular spatial framework that assumes the force of infection from one population to another decays with the distance between them, but the choice of distance metric can affect the predicted dynamics. Conventionally, great-circle distance is used, but humans don’t generally travel in a straight line. In this talk, we evaluate driving distance and driving time against great-circle distance as gravity model distance metrics by testing their ability to predict influenza dynamics in the US with fine-scale influenza-like-illness medical claims data. I will show that driving distance metrics can offer better model fits to infectious disease spread compared to great-circle distance, but that simulated predictions remain similar. However, the choice of distance metric can matter more depending on the terrain and the nature of the disease spread.