Statistical inference and planning for complex infectious disease systems

传染病对整个社会都很重要。大流行性H1N1流感、埃博拉病毒、SARS或口蹄疫等疾病的爆发对公共卫生构成直接威胁和/或产生严重的经济影响。这些疾病甚至可以被用作武器，作为生物或农业恐怖主义行为引入。为了控制这些疾病，必须了解疾病如何随着时间的推移而传播，以及哪些因素导致一些人感染，而另一些人则没有感染。为此，近年来已经开发了一系列用于传染病传播的“基于个体”或“个体水平”的数学模型。这种模型直观、灵活，并且已经被证明可以准确地描述先前观察到的流行病在空间和时间上的模式。至关重要的是，它们可以提供关于种群中个体的信息，如空间位置、工作地点、遗传信息等，要包含在模型中。这与经典的传染病传播模型形成了鲜明对比，后者假设人口中的每个人都是相同的，人口中的每个人与其他人接触的频率相同。统计推断是一个用于推导和评估模型的过程，这些模型由观察到的数据提供信息。将从真实的世界收集的数据合并到我们的模型中是有利的，因为它可以生成更好地反映现实的模型。因此，我们可以对从我们的模型得出的结论有更好的信心。计算密集型技术被用来执行这个推理过程。然而，当应用于这些复杂的个体水平模型时，这些密集型技术可能需要计算机很长时间才能执行，特别是如果我们有大量数据，大量缺失数据或我们希望解释的数据不准确;所有这些都是传染病数据集的典型情况。如果需要非常快的结果，这可能是一个主要问题。例如，我们可能希望决定如何在疾病爆发期间最好地控制疾病，因此需要尽快进行数据分析。因此，本研究的目的是：1）开发这些个人水平的模型，使它们能够更准确和可靠地模拟疾病传播，因为它发生在真实的生活（例如，通过使用病毒或细菌本身收集的遗传信息）; 2）改进计算密集型统计推断过程，使其更有效; 3）使用这些模型来帮助设计动物疾病传播实验，其目的是在受控环境中收集信息数据，以帮助了解疾病传播以及如何治疗和控制。这些发展可以用来进一步了解传染病，并通过这一点，我们控制它们或减轻不必要的严重后果的能力。

Infectious diseases are of great importance across society. Outbreaks of diseases such as pandemic H1N1 influenza, Ebola, SARS, or foot-and-mouth disease pose direct threats to public health and/or have serious economic effects. There diseases can even by used as weapons, introduced as acts of bio- or agro-terrorism.****In order to control such diseases, it is vital to understand how the disease spreads over time and what factors lead to some individuals becoming infected and some not. To this end, a series of mathematical 'individual-based' or 'individual-level' models for infectious disease transmission has been developed in recent years. Such models are intuitive, flexible, and have been shown to accurately describe the patterns of previously observed epidemics over space and time. Of key importance is the fact that they allow information about the individuals in the population, such as spatial location, where they work, their genetic information, etc., to be included in the model. This is in contrast to classic infectious disease transmission models that assume that everybody in the population is the same, and everybody in the population comes into contact with everybody else equally often.***Statistical inference is a process used to derive and assess models that are informed by observed data. It is advantageous to incorporate data collected from the real world into our models, as it produces models that better reflect reality. Thus, we can have better confidence in conclusions drawn from our models.***Computationally intensive techniques are used to carry out this inference process. However, when applied to these complex individual-level models, these intensive techniques can take the computer a long time to carry out, especially if we have a lot of data, lots of missing data, or inaccuracies in the data that we wish to account for; all are typically the case with infectious disease data sets. This can be a major problem if results are needed very quickly. For example, we may wish to decide how best to control a disease during the course of an outbreak, and so require data analysis to be carried out as quickly as possible.***The purpose of this research is therefore to do the following: 1) develop these individual-level models so they can more accurately and reliably model disease transmission as it occurs in real life (e.g., by using genetic information collected on the virus or bacteria itself); 2) improve the computationally intensive statistical inference process to make it more efficient; 3) use these models to help design animal disease transmission experiments, the purpose of which is to collect informative data in a controlled environment to aid understanding disease transmission and how it might be treated and controlled.***These developments can then be used to further our understanding of infectious diseases, and through this, our ability to control them or alleviate unnecessarily severe outcomes.**