Significance of multi-site calibration for agent-based transmission models

Simulation modeling has been leveraged for a variety of health care applications. Data used to inform most parameters for these models are typically collected from a single site, thus limiting the generalizability of the results and insights for other populations. In this study, we explore the impact of using data collected from multiple sites to parameterize an agent-based model of multidrug-resistant organisms in an intensive care unit setting. We show that using single sites to inform model parameters can be highly variable, and that using multiple sites can significantly improve the precision of these estimates and reduce the associated aggregated model error.