Mechanistic models, which incorporate feedback effects across different levels and over time, have been of increasing interest to the field of prevention science given their long history in infectious disease epidemiology. These models may be used to understand the drivers of new epidemics, and where to target new prevention strategies prior to implementation. Validated network models can be used to efficiently test hypotheses related to comparative intervention designs, policy effectiveness, and implementation strategies. In this symposium, we bring together an interdisciplinary set of prevention scientists that have used network models for HIV prevention to demonstrate the motivations for developing these models, latest methodological innovations, applications of their use in clinical and public health contexts, and best practices for validation and reproducibility.
The first paper, titled “Network Models for HIV/STI Transmission Dynamics: Statistical Methods and Computational Tools”, describes the historical background and motivation for the development of network modeling methods to meet the needs of a rapidly evolving HIV prevention landscape. It will describe the innovative, statistically principled approaches of modeling networks with Temporal Exponential-family Random Graph Models (TERGMs), with brief example applications on HIV prevention for men who have sex with men in the US.
The second paper, titled “Getting to Zero: Triangulating HIV incidence predictions from micro-level network modeling and macro-level incidence projections”, provides a comparison of macro-level methods to project the number of new HIV infections, with simulations of dynamic micro-level network structure. The network models use similar statistical theory that is foundational to the first paper in this compendium. This paper provides a triangulation of incidence projection using multiple methods, which will be used to inform the “Getting to Zero” campaign in Illinois.
The third paper, titled “The impact of the network formation process on HIV spreading behavior”, highlight the impact of often used network formation processes. It introduces an Agent-Based perspective into network formation and highlights the difference between ERGMs and Agent-based network formation models. It shows the impact of the formation process can have on model results by comparing the two, and their implications for prevention policy.
Our discussant, Prof. C. Hendricks Brown, will synergize the paper presentations highlighting the value of network simulation models, and lead presenters and attendees into a discussion on how to further integrate network simulation modeling into prevention science. We feel that exposure to various network simulation methodologies and examples, will be novel and appealing for the SPR Annual Meeting participants, and serve as a means to extend their prevention science toolbox.