Mechanistic models that predict disease rates based on intrinsic (i.e. those inherent to the epidemiological system, such as dynamics of immune response) or extrinsic (i.e. independent of the epidemiological system, such as climate change) can anticipate future patterns due to non-linear interactions that might not be apparent in future dynamics.  However, it is challenging to develop and parameterize these models to individual settings (e.g. because transmission and contact rates, or human behavior differ from place to place).  Forecasting has been more effectively used to predict changes in disease rates or invasion risk for vector-borne disease [e.g.Bhatt 2013], because the environmental conditions that are suitable for the vector can be reasonable studied in the absence of the pathogen.