Mathematical Analysis of serological data for Toxoplasma gondii in the human population, a Bayesian approach

Abstract

The dynamics of infections and their prevalence in populations are key elements for effective public health policies. In this study we investigated the change in prevalence and the force of infection over time within the Dutch population for Toxoplasma gondii, a widespread zoonotic foodborne parasite. We offer an innovative approach to modeling infection dynamics in disease surveillance, by applying Bayesian statistics and compartmental disease modeling. Using serological data from three independent studies conducted at 10-year intervals, a binary mixture model was implemented, characterizing the distribution of the measurements to estimate the prevalence without reliance on predetermined cut-off values for the classification of infected and non-infected subpopulations. Potential external covariates such as education level, pet ownership (specifically cats), and the consumption of food associated with higher risk of infection were explored to ascertain any changes in risk factors over time. A discrete-time and age-dependent SI(S) compartmental disease model was implemented to estimate the force of infection across years and ages.

This Master thesis project is conducted as part of an internship at the National Institute for Public Health and Environment (RIVM).