There are a variety of approaches to modeling and making predictions about the dynamics of infectious diseases. For instance, one can take a strategic/mechanistic approach that primarily concerns itself with determining what types of processes can cause observed patterns. Strategic models often require large amounts of data to parameterize them and make them useful for prediction. On the other extreme are tactical/phenomenological models, like regressions, that usually focus on fitting a pattern without elucidating why those patterns exist. Tactical models, while often conceptually simpler, can be poor for extrapolating beyond the range of the data. Thus each approach has it’s strengths and weaknesses in terms of data needed to parameterize and validate the model and the types of predictions that we can make using them. Here I present a phenomenological approach that uses Gaussian Processes (GPs) to forecast case data for infectious diseases. In particular I focus the dynamics of dengue, a mosquito-borne viral infection of humans as a case study.
Results/Conclusions
We find that GPs are a parsimonious and data-light approach to predicting the incidence of dengue. This approach performs better in terms of predictive accuracy than a more traditional time dependent GLM approach that includes environmental predictors of incidence as well as multiple other models of varying complexity for long time series of incidence at single locations.