Fitting dynamic models to epidemic outbreaks with quantified uncertainty: A primer for parameter uncertainty, identifiability, and forecasts

Chowell, Gerardo

In this paper we review and illustrate a simple data assimilation framework for connecting ordinary differential equation models to time series data describing the temporal progression of case counts relating to population growth or infectious disease transmission dynamics (e.g, daily incident cases). This frequentist approach relies on modeling the error structure in the data unlike Bayesian approaches which always raise the question of how to set priors for the parameters. We present examples based on phenomenological and mechanistic models of disease transmission dynamics together with simulated and real datasets. We discuss issues related to parameter identifiability, uncertainty quantification and propagation as well as model performance and forecasts.