RAPID: Optimal allocation of COVID-19 testing based on context-specific outbreak control objectives

该项目的目的是为目标驱动的、环境依赖的疾病监测战略设计一个框架，旨在处理抽样误差和偏差。该框架将应用于基于多种疫情情景的COVID-19检测分配，采用多种模型来改进COVID-19监测和控制的决策。这项工作将改善对COVID-19大流行的应对。结果将提交给主要联邦机构，以便将其作为COVID-19疫情决策过程的一部分加以考虑。这些方法也将高度适用于最佳疫苗分配，特别是在疫苗供应有限的早期阶段。此外，它还将为未来的疫情测试应对提供一个框架。培养1名博士后，学习应用流行病学研究的理论和方法。由于现有的不完善的检测方法越来越多，但数量仍然有限，分配检测方法的方式关键地决定了我们可以学到什么，进而决定了我们在为个人和群体管理疾病方面可以得出什么推论。这就提出了有限资源的最优分配问题。分配有限数量的测试的上下文依赖性质引入了额外的潜在误差和偏差来源，使最佳测试分配问题成为一个独特的挑战。目前的建模工作必然集中在疾病动力学和干预策略的有效性上，但很少有人明确考虑检测、接触者追踪和隔离策略。与原地避难或保持社交距离的规定不同，测试分配对管理战略决策的影响取决于密度。虽然测试仍然是有限的，但是显式地为测试分配和涉及监视和管理的策略建模是至关重要的。监测成功的关键是为特定目标积极设计监测战略。该项目将采用基于两个步骤的有效监测原则。首先，确定监测的目的。其次，调整抽样设计以实现这一目标，在这种情况下，以非代表性的方式选择个体组，并单独估计随机抽样的个体在这些组中出现的概率。由于不同检测的特异性和敏感性、不同检测平台的性能和人群水平的发病率，疾病状态的错误分类（例如假阳性/阴性）也将得到解决。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。

The purpose of this project is to design a framework for objective-driven, context dependent disease surveillance strategies, designed to deal with sampling errors and biases. This framework will be applied to the allocation of COVID-19 testing based on multiple outbreak scenarios, employing the use of multiple models to improve decision-making for COVID-19 surveillance and control. This work will improve the response to the COVID-19 pandemic. Results will be presented to key federal agencies, so that they can be considered as part of the decision-making process for the COVID-19 outbreak. These methods will also be highly applicable to optimal vaccine allocation especially during the early stages of vaccine availability when supplies will be limited. In addition, it will provide a framework for future outbreak testing response. One postdoctoral researcher will be trained in the theory and methods of applied epidemiological research. With a growing, but still limited number of imperfect tests available, the way in which tests are allocated critically determines what we can learn and in turn, what inferences can be made with respect to managing the disease for individuals and populations. This poses an optimal allocation problem for limited resources. The context-dependent nature of allocating a limited number of tests introduces additional potential sources of error and bias, making the question of optimal testing allocation a unique challenge. Current modeling efforts focus necessarily on disease dynamics and the efficacy of intervention strategies, but few consider testing, contact tracing, and isolation strategies explicitly. Unlike shelter-in-place or social distancing mandates, the impact of test allocation on management strategy decision-making is density-dependent. While testing remains limited, it is critical to explicitly model test allocation and strategies that involve both monitoring and management. The key to successful surveillance is to actively design surveillance strategies for a specific objective. The project will use a principle of effective monitoring based on two steps. First, identify the objective of the monitoring? Second, tailor the sampling design to achieve that objective, in this case selecting groups of individuals in a nonrepresentative way and to separately estimate the probabilities that a randomly sampled individual would appear in these groups. Misclassification of disease state (e.g., false positives/negatives) due to the specificity and sensitivity of different tests, the performance of different test platforms and population-level incidence will also be addressed.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.