Objectives matter: A mathematical modelling framework to identify optimal control strategies for future infectious disease outbreaks

在决定疫情期间的最佳控制干预措施时，政策制定者可能会寻求选择将成本、持续时间或严重疾病病例数量等已定义的流行病指标降至最低的战略。然而，最有可能的是，有关政策制定者被要求考虑多个相互竞争的目标，这使确定最佳战略变得复杂。不确定性使决策进一步复杂化，特别是如果病原体是新的，疾病特定参数或干预措施的效果未知。因此，政策制定者可以受益于一个框架，该框架可以根据包括多个目标的成本函数，为未来的传染病暴发提出控制战略。该项目旨在开发一个数学框架，该框架使用适当的目标函数来探索传染病暴发的控制干预措施的最优化。该项目将利用该框架调查最佳干预政策如何取决于疫情的时空状态、疾病的特征，以及至关重要的是，决策者在执行这些政策时的目标。这一框架将首先应用于具有全球意义的人类呼吸道病毒暴发模型，例如季节性流感。如果时间允许，框架将扩展到考虑可能需要替代建模方法和不同类别的控制干预措施的动植物疾病。该项目将依赖于各种传染病暴发模型的开发和模拟，这些模型的预测将被用来衡量目标函数。必须开发模型，以便灵活实施可行的控制干预措施，并使用统计推断将其与数据(由外部伙伴提供或通过以前的出版物和/或在线公共卫生仪表板公开获得)相适应，或纳入疾病参数中看似合理的不确定性，如传播率，以调查目标函数的变化。我们将开发一套模型，可用于一系列不同的未来传染病暴发，包括考虑随机模型和空间显式模型。就呼吸道疾病而言，考虑空间上显式的模型为建模增加了一层现实感，因为最近SARS-CoV-2(新冠肺炎)大流行期间的限制通常是局部的。这可以帮助政策制定者确定脆弱的行政区划并采取有针对性的行动。在植物和动物疾病的背景下，空间异质性对于模型的准确性至关重要，因为种群不再被认为是良好混合的。

When deciding upon the optimal control intervention during an epidemic, a policy maker might seek to select the strategy that minimises a defined epidemic metric such as cost, duration or the number of severe disease cases. It is most likely, however, that the policy maker in question is required to consider multiple competing objectives which complicates the identification of the optimal strategy. The decision is further complicated by uncertainty, especially if the pathogen is novel and disease-specific parameters or the efficacy of interventions are unknown. The policy maker could therefore benefit from a framework which can recommend control strategies for future infectious disease outbreaks, subject to a cost function which includes multiple objectives. This project aims to develop a mathematical framework which uses a suitable objective function to explore optimisation of control interventions for infectious disease outbreaks. The project will use the framework to investigate how optimal intervention policies are dependent upon the spatiotemporal state of the outbreak, the characteristics of the disease and, crucially, the objective of policy makers when implementing such policies. This framework will be initially applied to models of respiratory virus outbreaks in humans of global significance, such as seasonal influenza. Time permitting, the framework will be extended to consider animal or plant diseases which may require an alternate modelling approach and a different class of control interventions. The project will rely upon the development and simulation of various models ofinfectious disease outbreaks whose projections will be used to measure the objective function against. The models must be developed to allow for flexible implementation of feasible control interventions and will be fitted to data (either provided by the external partner or publicly available through previous publications and/or online public health dashboards) using statistical inference, or incorporate plausible uncertainty in disease parameters such as the transmission rate to investigate changes in the objective function. We will develop a suite of models that can be used in a range of different futureinfectious disease outbreaks, including considering stochastic models and spatially explicit models. In the case of respiratory diseases, considering a spatially explicit model adds a layer of realism to the modelling as restrictions during the recent SARS-CoV-2 (COVID-19) pandemic were often localised. This can assist the policy maker in identifying vulnerable administrative divisions and taking targeted action. In the context of plant and animal diseases, the spatial heterogeneities are essential for model accuracy as the population can no longer be considered to be well-mixed.