Dissertation Title:  "Empirical Approaches to Modeling HIV and Hepatitis C"This thesis describes applications of mathematical models to three distinct problems. The common objective of the three papers was to take full advantage of all available data sources in constructing and fitting models of disease processes, their impact at the population level or the potential costs and benefits of interventions.

The goal of Chapter 1 was to extend previous models of HIV/AIDS epidemics in sub-Saharan Africa in order to improve the methodological basis for epidemiological estimates given limited available data sources. A simple model was developed in which past trends in HIV incidence were back-calculated from data on the prevalence of infection in adults. The model was then used to develop estimates and ranges of uncertainty around HIV/AIDS epidemics in four countries in sub-Saharan Africa.

Chapter 2 describes a more complex model of the hepatitis C virus (HCV) epidemic in the United States. The objectives of the study were to develop a comprehensive model that was consistent with both the current understanding of the natural history of the disease and the observed impact of the disease at the population level. Extensive model-fitting was undertaken through multiple simulations using model parameter values sampled from plausible ranges. The parameter values that were consistent with observed data on prevalence and mortality were identified and used to explore uncertainties about the natural history and epidemiology of HCV.

Chapter 3 presents a re-examination of the cost-effectiveness of treatment for HCV. A series of analyses were undertaken in order to examine the impact of critical modeling assumptions about disease progression rates and health-related quality of life in HCV-infected patients on conclusions about the potential costs and benefits of different treatment strategies. The results of this study suggest that key uncertainties around the prognosis and health experience of HCV patients may be inadequately reflected in conventional sensitivity analyses, but may have important implications for the clinical management of individual patients and for broader health policy decisions.