Representing contact heterogeneity: Scaling up from a partially observed network

Multiple lines of evidence suggest that contact heterogeneity may be important for understanding disease spread.  However, contact networks are often only partially observable.  We discuss challenges with and a method for scaling partial data to estimate a full network.  We use partially observed data on a livestock shipment network for the US to estimate a contact network and drive a disease model for introduction of foreign animal disease in the US.  We infer the relative importance of long distance movement across the network to local spread and discuss implications for control.