Dynamics, Underlying Mechanisms, and Uncertainty in Environmentally Driven Infectious Diseases

Abstract

The inherent complexity of environmentally driven infectious disease dynamics poses both methodological and practical challenges for epidemiological research. In this dissertation, we use a combination of time series analysis and mathematical modeling to examine these complex dynamics and their underlying mechanisms, in two different disease contexts. We also explore issues of uncertainty that arise when using these methods to study disease transmission and interventions. First, we formally evaluated both the structural and practical identifiability of a common transmission model of mosquito-borne diseases, using the 2010 dengue epidemic in Taiwan as a case study. We found that while the model is structurally identifiable, it is practically unidentifiable under a range of human and mosquito time series measurement scenarios, potentially leading to incorrect predictions of the effects of interventions. However, the basic reproduction number (R0) remains stable across different parameter values in spite of the unidentifiability of the individual parameters. These identifiability issues can be resolved by directly measuring additional human and mosquito life-cycle parameters both experimentally and in the field. This work illustrates the importance of examining identifiability when linking models with data to make predictions and inferences, and particularly highlights the importance of combining laboratory, field, and surveillance data to successfully estimate epidemiological and ecological parameters using models. Second, we revisited the first two waves of the 2010 Haiti cholera epidemic to assess the potential benefits of integrating mobile-phone data into disease transmission models, and to examine the relative roles of rainfall and human movement in the early dynamics. Using a susceptible-infectious-water-recovered (SIWR) framework, we compared models integrating mobile-phone-based movement, gravity-model-based movement, and no movement, either with or without rainfall. We found that both gravity and mobile-phone based movement model can capture the timing of cholera introduction into each department. Once introduced, rainfall becomes the major mechanism to drive the cholera epidemics. By better understanding these early dynamics, this work provides insights for both the ongoing epidemic in Haiti, as well as other emerging outbreaks. Finally, we characterized the seasonality and the temporal associations between cholera incidence, rainfall, and temperature in the subsequent cholera epidemics in Haiti following the first two epidemic waves. Using wavelet analysis and distributed lag nonlinear models (DLNMs), we found that temperature is not an informative predictor of cholera incidence. Rainfall, on the other hand, remains a seasonal driver of cholera incidence with lags between rainfall and cases increasing over time; however, rainfall alone is still not sufficient to explain all the features observed in the disease dynamics. Our analysis also showed that different departments have a distinct pattern for the interaction between rainfall and cholera, which highlights the importance of studies with finer resolutions and the potential for developing targeted, regional intervention strategies. The results of this dissertation highlight the importance of integrating a wide range of data sources and methodological approaches. Through careful consideration of the range of potential mechanisms and sources of uncertainty, we show that computational models can be used to better understand the roles and relative importance of these mechanisms in the spread of environmentally driven diseases. we also discuss potential ways these findings could be translated into practical implications and policies for disease interventions and elimination.