Dynamic modelling of foot-and-mouth disease (FMD) epidemiology and persistence in endemic areas

越来越清楚的是，为了了解病原体的生态学和流行病学，我们还需要了解它们的进化并将其纳入模型。这对于快速进化的病毒（如动物的口蹄疫或人类的流感）尤其重要（Roche，Drake和Rohani，2011）。这一点很重要，因为假设随机混合和固定病原体特征的简单微分方程模型可能有偏差，并高估了疫苗接种和行动限制等控制措施的影响。此外，大多数当前的动物疾病模型很大程度上是在群体水平上定义的，很少包括群体动态或解释非洲等公共牧场中出现的自由放养运动。因此，为了了解疾病控制如何在大的地理范围内发挥作用，例如在撒哈拉以南非洲（SSA），重要的是要了解病原体如何进化以逃避宿主的免疫系统，以及如何与宿主免疫力减弱或通过出生在种群中积累新的幼稚动物交叉。当试图大规模根除疾病时，理解这些影响将变得至关重要，盖茨基金会、粮农组织和世界动物卫生组织目前正在考虑根除口蹄疫（Kitching等人，2007年）。与天花和牛瘟不同，在天花和牛瘟中，存在对所有菌株产生终身免疫力的单一疫苗，在口蹄疫的情况下，存在多种血清型（目前已知在SSA中传播的6种），并且菌株和当前疫苗诱导短暂的免疫力，很少或没有交叉保护，导致需要定期疫苗匹配。流行病学和进化时间尺度之间的重叠，在牲畜种群中观察到的免疫漂移，以及RNA病毒的高突变率，使得将大的菌株空间整合到模型中至关重要。因此，经典的建模方法，熟悉的SIR框架的基础上，不容易服从于此目的，因为由此产生的状态空间的应变的数量迅速增加，因此，变得过于繁琐的有意义的分析。因此，以FMD为例，本项目将开发一个基于代理的建模框架。这种建模的基本原理是探索个体异质性，使我们能够跟踪感染历史。该框架将包括覆盖在传播模型上的运动网络和病毒进化层，以在计算机上探索这些动态，从而开发可以探索特定控制方案的理论基础。该项目将以口蹄疫为例，重点研究RNA病毒流行病学中两个关键且相互关联的问题：i）病毒在地方性环境中持续存在的规模; ii）持续存在的驱动因素（本地或远程牲畜移动、病毒进化、宿主免疫力减弱、种群规模、携带状态等）是什么？该项目借鉴了主管人员在非洲从事口蹄疫工作和建立疾病传播模型的长期经验（Bronsvoort，2003年）。此外，我们还在喀麦隆开展工作，在那里开展了大部分关于地方性口蹄疫的原始工作，描述了牲畜通过市场的流动情况，并使用GPS项圈在牛群和季节性迁移之间绘制地图。以前的系统发育分析FMD传播的空间传播和进化变化的病毒分析，但没有明确纳入这些考虑到当地的传播。此外，最近的工作，确定当过去暴露于感染将保护对新出现的菌株可能是至关重要的病毒持久性在一定程度上，但仍有待充分理解，并纳入病毒传播的模型。

Increasingly it is clear that to understand the ecology and epidemiology of pathogens we also need to understand and incorporate their evolution in models. This is particularly critical for rapidly evolving viruses such as foot-and-mouth disease in animals or influenza in humans (Roche, Drake, & Rohani, 2011). This is important because simple differential equation models that assume random mixing and fixed pathogen characteristics are likely to be biased and over estimate the impact of controls such as vaccination and movement restrictions. Furthermore, most current animal disease models are largely defined at a herd level and rarely include within herd dynamics or account for the free ranging movements seen in communal grazing areas such as in Africa. Therefore to understand how disease control might work at large geographical scales such as in sub-Saharan Africa (SSA) it is important to understand the how pathogens are evolving to evade the immune system of a host and how that intersects with host immunity waning or the accumulation of new naïve animals in the population through births. Understanding of these effects will become essential when attempting large scale disease eradication as currently being considered by the Gates Foundation, FAO and OIE for foot-and-mouth disease (Kitching et al., 2007). Unlike smallpox and rinderpest, where there was a single vaccine that produced life long immunity against all strains, in the case of foot-and-mouth disease there are multiple serotypes (6 currently known to circulate in SSA) and strains and current vaccines induce short lived immunity with little or no cross protection resulting in the need for regular vaccine matching. The overlap between epidemiological and evolutionary time scales, the observed immune drift in the livestock populations, and the high mutation rate of RNA viruses make it essential to integrate a large strain space into models. Consequently, the classic modeling approaches, based on the familiar SIR framework, are not readily amenable for this purpose because the resulting state space increases rapidly with the number of strains and, consequently, becomes too cumbersome for meaningful analysis. Therefore using FMD as an example this project will develop an agent based modelling framework. The rationale of this kind of modelling is to explore individual heterogeneity, allowing us to track infection history. This framework will include movement networks and virus evolution layers over laid on the transmission model to explore these dynamics in silico to develop the theoretical bases from which specific control options can be explored. These type of models are now possible because of the availability of cheap processing power.The project will focus on two key and interlinked questions in RNA viral epidemiology using FMD as the exemplar: i) at what scale(s) do viruses persist in an endemic setting and ii) what are the drivers of persistence (local or long-range livestock movement, virus evolution, waning host immunity, population size, carrier state etc)? The project builds on the supervisors long experience of working on FMD in Africa and in modelling disease spread (Bronsvoort, 2003). In addition we have on going work in Cameroon where much of the original work on endemic FMD was carried out, describing livestock movements through markets and using GPS collars to map within, between herd and transhumance movements. Previous phylogenetic analyses of FMD spread have informed analyses of the spatial spread of the virus and evolutionary changes but have not explicitly incorporated these in consideration of local spread. Furthermore, recent work on identifying when past exposure to an infection will protect against newly emerging strains is likely to be critical to viral persistence at some scale but remains to be fully understood and incorporated in models of viral spread.