III: Small: Data-Driven Control of Epidemic Processes over Complex Dynamic Networks

尽管上个世纪医学取得了显著进步，但最近的COVID-19等大流行提醒我们，传染病对人类的威胁是非常真实的。虽然医学的不断进步是必不可少的，但信息技术可以大大提高我们检测和控制传染病破坏性影响的能力。在这方面，公共卫生机构收集、定期更新并公开报告现场数据，其中包含受疾病影响地区的检测、感染、康复、住院和死亡人员的地理位置信息。然而，这些数据是不可靠的、不完整的和粗粒度的；因此，卫生机构可以极大地受益于信息技术来过滤和分析现场数据，以便对该疾病未来的传播作出可靠的预测。此外，卫生机构的最终目标是利用这些信息设计有效的战略，以遏制传染病的传播。为了实现这一目标，卫生机构拥有流行病控制资源，例如保持社会距离、交通限制和分配药物资源（只要有）。由于这些资源的异质性和高成本，在整个人口中找到每种资源的成本最优配置是一个具有最大社会影响的非常具有挑战性的问题。在这个项目中，我们建议开发一个综合框架，用于利用有限的资源和不可靠的数据对流行病爆发进行建模、预测和成本最优控制。为了实施实用的流行病控制工具，必须首先建立能够复制疾病传播的显著地理-时间特征的数学模型。这些模式受到疾病蔓延地区的地理以及人口流动模式的强烈影响。在这个方向上，我们将使用复杂的接触图来模拟现实的地理约束和移动模式。特别是，该图的顶点对应于城镇/地区，其链接表示它们之间的相互作用。在这个接触图的顶部，我们将建立一个动态模型，旨在复制疾病复杂的地理-时间传播。在这个方向上，我们将考虑一个随机过程系统，通过接触图的边缘耦合，来模拟疾病的演变。一旦调整了传播模型，我们就会着手设计一种协调的策略，通过在人口中分配资源来控制感染的传播。在这个方向上，我们将设计并实现一个优化程序，以找到在有限预算下异构资源的成本最优分配。在这项研究任务中，我们必须处理现场数据固有的不确定性，以及可能对成本最优资源分配的公平性产生巨大影响的抽样偏差的存在。拟议研究计划的成功将大大提高我们有效发现和适当应对流行病爆发的能力，从而可以部署快速控制反应。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Despite notable advances in medicine over the last century, recent pandemics such as COVID-19 remind us that the threat of infectious diseases to human populations is very real. While continuing advances in medicine are essential, information technologies can greatly improve our ability to detect and contain the devastating effects of infectious diseases. In this direction, public health agencies collect, periodically update, and publicly report field data containing geolocated information about the tested, infected, recovered, hospitalized, and deceased individuals in those areas affected by the disease. However, this data is unreliable, incomplete, and coarse-grained; therefore, health agencies can greatly benefit from information technologies to filter and analyze field data in order to make reliable predictions about the future spread of the disease. Moreover, the final objective of a health agency is to use this information to design efficient strategies to contain the spread of infectious diseases. To achieve this objective, health agencies have at their disposal epidemic-control resources, such as social distancing, traffic restrictions, and the distribution of pharmaceutical resources (whenever available). Due to the heterogeneity and high cost of these resources, finding the cost-optimal allocation of each type of resource throughout the population is a very challenging problem of utmost societal impact. In this project, we propose to develop an integrated framework for modeling, prediction, and cost-optimal control of epidemic outbreaks using finite resources and unreliable data.In order to implement practical epidemic-control tools, it is necessary to first develop mathematical models able to replicate salient geo-temporal features of disease transmission. These patterns are strongly influenced by the geography of the area over which the disease is spreading, as well as the mobility patterns of the population. In this direction, we will use complex contact graphs to model both realistic geographical constraints and mobility patterns. In particular, the vertices of this graph correspond to towns/districts and its links represent interactions between them. On top of this contact graph, we will build a dynamical model aiming to replicate the complex geo-temporal spread of the disease. In this direction, we will consider a system of stochastic processes, coupled through the edges of the contact graph, to model the evolution of the disease. Once the model of the spread is tuned, we will then proceed to the design of a coordinated strategy to contain the spread of the infection by distributing resources throughout the population. In this direction, we will design and implement an optimization program to find the cost-optimal allocation of heterogeneous resources given a finite budget. In this research task, we must deal with the inherent uncertainty of field data, as well as the presence of sampling biases that can have a dramatic impact on the fairness of the cost-optimal allocation of resources. The success of the proposed research program would greatly improve our ability to efficiently detect and appropriately react to epidemic outbreaks, whereupon a rapid control response can be deployed.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.