A Novel Approach to Modeling Zika Case Counts for Demographic Groups

Abstract

Zika, a formerly obscure, little-studied virus, recently burst onto the global scene, infecting over a million individuals across more than 33 countries in under a year. The virus ignited an explosion of research that focused on simulating weekly case counts through mathematical models and characterizing the two most potent consequences of the disease, microcephaly and Guillain-Barré Syndrome (GBS). While such models have largely been successful, little literature has attempted to develop predictions for case counts by demographic group. This is an especially important deficiency given preliminary analyses suggesting that Zika may disproportionately affect certain demographic groups, notably women. This thesis aims to fill this gap by creating a novel, flexible framework to generate weekly demographic case counts using simple demographic data.We focus on analyzing gender and population density in the 2013-2014 outbreak of Zika in French Polynesia. To begin, we survey existing literature to select a base framework, design a noisy case assignment policy, and identify a series of transmission parameter modifications. Next, we implement the assignment policy and the parameter modifications in the base framework, validate the predictions using self-developed evaluation criterion, and contextualize the results in broader literature. Finally, we conduct an illustrative impact analysis to demonstrate how our model can be used to guide policymakers and public health officials. We end with a discussion of our methodology’s limitations, potential for future work, and closing remarks.This thesis establishes the ability for our model to predict weekly case counts by gender in French Polynesia and, more broadly, leverage population-level demographic proportions to derive a reasonable range of case counts, opening the door to future characterization and prediction built upon the framework outlined in this paper.