The Structure of an Epidemic: A Model of AIDS Transmission in Sub-Saharan Africa

Patrick Heuveline, University of Chicago
David Sallach, University of Chicago
Thomas Howe, University of Chicago

Agent-based models can address some of the shortcomings of present macro bio-behavioral models of the HIV epidemic that segregate the population into behavior-based "risk groups." Modeling social network allows these models to assess individual risk heterogeneity within "risk groups," that is, why the risk associated with a given behavior varies, while modeling the social processes that contribute to such behaviors allows them to represent how such groups are formed, maintained and dispersed. This paper describes the development of an interactive computational model based upon relatively simple assumptions. The model begins by representing an individual's history of employment, migration, birth, death, courtship, marriage, as well as her family, affinity, and sexual networks. Data underlying the relevant rates and assumptions are drawn from a variety of sources, and are parameterized to allow an assessment of the impact of diverse assumptions as well as to illustrate interaction effects.