Abstract
The dynamics of infectious diseases are greatly influenced by the movement of both susceptible and infected hosts. To accurately represent disease dynamics among a mobile host population, detailed movement models have been coupled with disease transmission models. However, a number of different host movement models have been proposed, each with their own set of assumptions and results that differ from the other models. Here, we compare two movement models coupled to the same disease transmission model using network analyses. This application of network analysis allows us to evaluate the fit and accuracy of the movement model in a multilevel modeling framework with more detail than established statistical modeling fitting methods. We used data that detailed mobile pastoralists' movements as input for 100 stochastic simulations of a Spatio-Temporal Movement (STM) model and 100 stochastic simulations of an Individual Movement Model (IMM). Both models represent dynamic movement and subsequent contacts. We generated networks in which nodes represent camps and edges represent the distance between camps. We simulated pathogen transmission over these networks and tested five network metrics-strength, betweenness centrality, three-step reach, density, and transitivity-to determine which could predict disease simulation outcomes and thereby be used to correlate model simulation results with disease transmission simulations. We found that strength, network density, and three-step reach of movement model results correlated with the final epidemic size of outbreak simulations. Betweenness centrality only weakly correlated for the IMM model. Transitivity only weakly correlated for the STM model and time-varying IMM model metrics. We conclude that movement models coupled with disease transmission models can affect disease transmission results and should be carefully considered and vetted when modeling pathogen spread in mobile host populations. Strength, network density, and three-step reach can be used to evaluate movement models before disease simulations to predict final outbreak sizes. These findings can contribute to the analysis of multilevel models across systems.
Original language | English |
---|---|
Article number | e1007184 |
Journal | PLoS Computational Biology |
Volume | 15 |
Issue number | 8 |
DOIs | |
Publication status | Published - 2019 |
Externally published | Yes |
Bibliographical note
Publisher Copyright:Copyright: © 2019 Pomeroy et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.