Abstract

Progress in experimental techniques provides new insight into the dynamics of infection within hosts. In order to make inference on the processes driving the pathogen’s population dynamics and assess the validity of biological hypotheses, it is necessary to design mechanistic mathematical models which can be fitted to the data. The objectives are usually to compare alternative models and estimate the parameters of the “best-fitting model(s)”. The last few years have seen rapid development in the statistical frameworks available, but there remain issues with their ability to deal with either complex stochastic models or with uncertainty in the data collection process. I will present my ongoing attempt to address these two issues within a classic maximum-likelihood framework, in the hope that it can be improved or complemented by some state-of-the-art Bayesian algorithms.