Department of Applied Mathematics and Statistics, Colorado School of Mines, Golden, CO, USA; Information Systems and Modeling Group (A-1), Analytics, ALD Global Security, Los Alamos National Laboratory, Los Alamos, NM, USA.
Department of Biostatistics, University of Iowa, Iowa City, IA, USA.
Spat Spatiotemporal Epidemiol. 2022 Jun;41:100505. doi: 10.1016/j.sste.2022.100505. Epub 2022 Apr 7.
The dynamics of human infectious diseases are challenging to understand, particularly when a pathogen spreads spatially over a large region. We present a stochastic, spatially-heterogeneous model framework derived from the foundational SEIR compartmental model. These models utilize a graph structure of spatial locations, facilitating mobility via random walks while progressing through disease states, parameterized by the net probability flux between locations. The analysis is bolstered by Approximate Bayesian Computation, by which epidemiological and mobility parameter distributions are estimated, including an empirically adjusted reproductive number, while model structure proposals are compared using Bayes Factors. The utility of this novel class of models is demonstrated through application to the 2014-2016 Ebola outbreak in West Africa. The flexibility of such models, whose complexity may be adjusted as desired, and complementary methods of analysis enable the exploration of various spatial divisions and mobility schema, while maintaining the essential spatiotemporal disease dynamics.
人类传染病的动态变化难以理解，特别是当病原体在一个大区域内空间传播时。我们提出了一个由基础 SEIR compartmental 模型推导而来的随机、空间异质模型框架。这些模型利用空间位置的图结构，通过随机游走促进移动性，同时通过位置之间的净概率通量来逐步发展疾病状态。近似贝叶斯计算（Approximate Bayesian Computation）用于对疾病传播和移动参数分布进行估计，包括对经验调整后的繁殖数的估计，同时使用贝叶斯因子（Bayes Factors）来比较模型结构建议。通过应用于 2014-2016 年西非的埃博拉疫情爆发，证明了这种新型模型的实用性。这种模型的灵活性，其复杂性可以根据需要进行调整，以及补充分析方法，使得可以探索各种空间划分和移动模式，同时保持基本的时空疾病动态。
