Thomas J. Hladish – University of Florida
Abstract: Dengue vaccines will soon provide a new tool for controlling dengue transmission, but the effectiveness of widespread, long-term vaccination campaigns remains uncertain. We developed an agent-based dengue model that describes the movements and transmission dynamics of people and infected mosquitoes in Yucatan, Mexico, and evaluated the effectiveness of various vaccine deployment strategies via simulation. Because the long-term durability of vaccine-induced immunity is unknown, we consider two different vaccine mechanisms. We also examined the effectiveness of combining vaccine campaigns with increased vector control, and the consequences of stopping vector control. When possible, we designed the model to use data sources with international coverage, to simplify re-parameterization for other regions. Our approach integrates satellite imagery, census and economic data, and 35 years of dengue caseand serotype data to model the locations and movement of 1.8 million people in 375,000 households and 100,000 workplaces and schools. In order to fitcertain important parameters that are poorly known, like the number of mosquitoes in Yucatan and their movement and biting behavior, we have developed a Bayesian parameter estimation toolkit in C++ called AbcSmc (Approximate Bayesian Computation – Sequential Monte Carlo). We present the overall effectiveness (i.e.,reduction in cases and hospitalizations relative to a non-intervention scenario) of several vaccination campaign and vector control strategies over a 20 year period. We find that plausible vaccination or vector control scenarios used separately may have high effectiveness (>75%) within the first five
years, but equilibrium effectiveness is modest (<50%). Furthermore, naive approaches to intervention (e.g., ceasing controls when they appear to have lost effectiveness) can result in years that are worse than what would be expected had no intervention happened at all, though in general there are still cumulative benefits when the intervention is considered over several years. Although the best-case scenario results in short and long-term effectiveness >90%, this would require both durable vaccine-induced immunity and a consistent, well-managed vector control program.
Short bio: Thomas J. Hladish is a research scientist at the University of Florida in the Department of Biology and the Emerging Pathogens Institute. Tom studies the temporal and spatial dynamics of infectious diseases. His recent work has focused on dengue, particularly using models to understand why the disease has proven difficult to control, and how we should best use interventions that may only be partially effective. He has developed several software projects to facilitate research in computational epidemiology, including AbcSmc, a toolkit for parallelized Bayesian parameter inference of stochastic models, and EpiFire, a contact network-based library for constructing populations and simulating disease transmission. Tom is also highly interested in education, particularly in helping the public health community understand how to use and interpret mathematical models of disease dynamics. For the past eight years he has been an instructor for the Summer Institute in Statistics and Modeling in Infectious Diseases at the University of Washington. While at the University of Texas, he worked with Lauren Ancel Meyers to develop and teach the first computational biology course at the university. He has taught courses in Ghana, Cuba, and Peru on various subjects in epidemiology and software engineering.
Local: Sala 2077 do ICEX