Using Agent Based Modelling and Mobility Data to Predict and Respond to the Outbreak and Spread of Infectious Diseases: A Case Study of SARS-CoV-2 in

研究背景SARS-CoV-2大流行继续摧毁卫生系统和经济，给世界上最脆弱的人群带来最大的风险。在编写本报告时，全世界已有超过2.03亿例病例和430万例报告死亡。这些后果对低收入和中等收入国家构成了特别严重的威胁，由于社会混合模式和大家庭规模，这些国家可能有更高的广泛传播可能性。此外，由于严重缺乏重症监护能力，他们面对大流行的准备最少(Walker等人)。2020)。世界银行最近估计，新冠肺炎可能在2020年内使7,100万人陷入极端贫困(Mahler等人，2020)。目的本研究的目的有三个方面：1.回顾现有的疾病建模方法，以及基于代理的建模所具有的潜力。建立基于人口数据、危险因素和流动数据的智能体模型，预测SARS-CoV-2和COVID-193疫情的时空传播和传播轨迹。基于空间建模和行为研究的结合，为应对未来的疾病暴发和快速遏制方法提出政策建议这项研究将首先审查迄今为止基于代理的建模用于疾病建模的方法，包括但不限于SARS-CoV-2。它还将评估作业成本模型与更传统的基于方程的模型的相对优点，并查阅有关新冠肺炎已知特征的文献。代理在模型中移动的方式将由移动数据提供信息。由于正在进行的工作，研究人员可以访问概述的所有类型的数据源。数据点在(1)传播风险或(2)严重疾病风险方面为模型提供信息。案例数据在相关地区内播种，然后模拟模型设置为运行。这使我们能够了解该国每个地区何时何地面临最大的传播风险。尽管该模型将能够告诉我们人们在应对疫情时的行为，但它不会告诉我们为什么。因此，为了为最终政策建议的设计提供信息，研究人员将在之前研究的基础上，进行实地工作，以更好地了解职业和年龄段之间的社会联系网络的差异。政策建议将寻求解决有关封锁和隔离的问题，但尽可能有更本地化的目标。研究的贡献由于持续的参与，这项研究将立即向津巴布韦国家卫生研究所和医学研究委员会提供成果。一旦开发出来，该方法就有可能应用于其他传染病的爆发。这项工作将寻求解决流动性数据能够在多大程度上解释传染病传播的问题；突出ABM方法的相对效用；并确定对疾病暴发应用更本地化的政策应对的优点。

Background of the research The SARS-CoV-2 pandemic continues to devastate health systems and economies, placing the largest risk on the world's most vulnerable populations. At the time of writing, there have been over 203 million cases worldwide and 4.3 million reported deaths (JHU). These consequences pose a particularly serious threat to Low and Middle Income Countries (LMICs), which may have higher likelihood of widespread transmission due to social mixing patterns and large household sizes. Furthermore, they are least prepared to face the pandemic due to severe lack of intensive care capacity (Walker et al. 2020). The World Bank recently estimated that COVID-19 could push 71 million into extreme poverty within 2020 (Mahler et al, 2020). AimsThe aims of this study are three fold: 1. Review existing approaches to disease modelling, and the potential that Agent Based Modelling holds2. Build an Agent Based Model informed by population data, risk factors and mobility data to predict the spatial and temporal spread and trajectory of SARS-CoV-2 and the risk of COVID-193. Based on a combination of spatial modelling and behavioural research, propose policy recommendations for tackling future disease outbreaks and rapid containmentMethodologyThe research will begin by reviewing the ways in which Agent Based Modelling has been used thus far for disease modelling, including but not limited to SARS-CoV-2. It will also assess the relative merits of ABM versus more traditional Equation Based Models and survey the literature on the known characteristics of COVID-19. The way the agents move around the model will be informed by mobility data. The researcher has access to all types of data sources outlined due to ongoing work. Data points inform the models in either (1) risk of transmission or (2) risk of severe disease. Case data is seeded within the relevant districts and then the simulation model is set to run. This enables us to understand when and where each part of the country is at greatest risk of transmission. Though the model will be able to tell us how people are behaving in response to the outbreak, it will not tell us why. Therefore, to inform the design of the final policy recommendations, the researcher will conduct fieldwork to better understand how social contact networks differ between occupation and age groups, building on previous research. Policy recommendations will seek to address the questions around lockdowns and quarantines but with a more localized targeting wherever possible. Contribution of ResearchThe research will provide immediate outputs to the National Institute of Health Research and the Medical Research Council's of Zimbabwe due to an on-going engagement. Once developed, the methodology has scope for application to other infectious disease outbreaks. The work will seek to address the extent to which mobility data can explain the spread of infectious disease; highlight the relative utility of ABM approaches; and identify the merits of applying more localized policy responses to disease outbreaks.