Angevaare Justin, Feng Zeny, Deardon Rob
University of Guelph, Canada. Electronic address: https://jangevaare.github.io.
University of Guelph, Canada. Electronic address: https://zfeng.uoguelph.ca.
Spat Spatiotemporal Epidemiol. 2021 Jun;37:100410. doi: 10.1016/j.sste.2021.100410. Epub 2021 Jan 29.
Transmission networks indicate who-infected-whom in epidemics. Reconstruction of transmission networks is invaluable in applying and developing effective control strategies for infectious diseases. We introduce transmission network individual level models (TN-ILMs), a competing-risk, continuous time extension to individual level model framework for infectious diseases of Deardon et al. (2010). Through simulation study using a Julia language software package, Pathogen.jl, we explore the models with respect to their ability to jointly infer latent event times, latent disease transmission networks, and the TN-ILM parameters. We find good parameter, event time, and transmission network inference, with enhanced performance for inference of transmission networks in epidemic simulations that have higher spatial signals in their infectivity kernel. Finally, an application of a TN-ILM to data from a greenhouse experiment on the spread of tomato spotted wilt virus is presented.
传播网络显示了传染病流行中的感染关系。重建传播网络对于应用和制定有效的传染病控制策略至关重要。我们引入了传播网络个体水平模型（TN-ILMs），它是对Deardon等人（2010年）的传染病个体水平模型框架的一种竞争风险连续时间扩展。通过使用Julia语言软件包Pathogen.jl进行模拟研究，我们探讨了这些模型在联合推断潜在事件时间、潜在疾病传播网络和TN-ILM参数方面的能力。我们发现这些模型在参数、事件时间和传播网络推断方面表现良好，在感染内核中具有较高空间信号的疫情模拟中，传播网络的推断性能得到了增强。最后，展示了TN-ILM在温室番茄斑萎病毒传播实验数据中的应用。
