School of Public Health

Samuel Jenness

Stochastic network models for HIV-1 transmission dynamics

The studies in this dissertation investigate how the structure of dynamically evolving sexual networks shape the HIV-1 epidemic among heterosexuals in Sub-Saharan Africa (SSA). The aims of this project were to: 1) develop a comprehensive set of demographic, behavioral, and biological parameters characterizing the target population of heterosexuals in SSA for use in network-based mathematical models of HIV transmission dynamics; 2) mathematically model the synergistic effects of network structure and male circumcision on HIV transmission in SSA; and 3) use mathematical modeling to estimate the total, direct, and indirect effects of pre-exposure prophylaxis (PreP) on HIV incidence within a simulated randomized control trial (RCT) environment with counterfactual scenarios for threshold levels of network structures that can bias the estimation of treatment efficacy. The heterosexual spread of HIV-1 infection in SSA depends on the unique configurations of how sexually active persons form and break sexual partnerships over time. For disease transmission, individuals are linked in dyads through partnerships, dyads are connected to other dyads when persons have multiple ongoing partnerships, and this forms the larger sexual network within the population. Effective HIV prevention tools, like male circumcision and PreP, operate within this network context. The studies here ask how these interventions, targeted at individuals, function given dynamic networks. Stochastic network models were developed to test key hypotheses for this interaction, aimed both at the population level and within RCT settings. Parameters for these models were primarily drawn from an original retrospective panel study we conducted in Accra, Ghana specifically for mathematical modeling. The findings from these studies address important empirical questions on the relationship between biomedical prevention tools and socio-behavioral risk, and also provide insight into the design and targeting of single-element and combination prevention packages for HIV in high-incidence settings. The broader mathematical modeling methods in this project have many potential applications for future HIV prevention research.