Spatio-temporal data integration methods for infectious disease surveillance

项目概要/摘要 有效的监测系统对于制定有针对性的、有效的公共卫生对策至关重要， 传染病在许多国内和全球环境中，正在采用多种系统， 重叠程度不同，每个系统旨在实现不同的监测目标。的目标 该研究项目是开发和分发一个时空数据集成工具集（OPTI-SURVEIL）， 分析来自多个监控系统的数据，通过跨系统链接数据来减少空间不确定性， 空间和时间在美国疾控中心和中国疾控中心的支持和指导下（见支持信），我们将 开发实用的多系统分析工具，将：（1）提供对各地区疾病负担的更好估计; 空间和时间;（2）评估个案确定中个人和系统层面的敏感性和偏差;（3） 查明目前监测战略中的冗余和差距。方法：在目标1中，我们将开发空间- 临时捕获-再捕获方法，用于在个人级别执行数据集成。述方法将 在病例确定和关键系统级参数（灵敏度， 偏差、依赖性）。这些方法将扩展到同时考虑多种疾病，以便 解决对合并感染负担的日益关注。在目标2中，我们将开发时空贝叶斯 用于使用通常由以下各项产生的聚合监视数据来执行数据集成的分层模型： 公共监控数据库。该模型将利用县级情况下的时空依赖性 数字，说明由于系统可用性而导致的时空缺失数据，并概率性地将 关于个人和系统层面的确定敏感性和偏倚的信息。在目标3中，我们 开发基于模拟的方法，对多种监视系统设计进行联合评估。这些 方法将应用于优化系统的设计，以最大限度地提高案件检测，同时考虑资源 约束、系统灵敏度/偏差以及其他系统的存在。我们将应用OPTIM-SURVEIL 中国现有的数据，并将重点关注四种全球重要的传染病：结核病（TB），疟疾， 血吸虫病和钩虫这些疾病表现出各种各样的监测挑战，包括 诊断准确性，随着消除的临近，越来越罕见的病例计数，疾病严重程度的可变性， 以及识别关键合并感染的挑战（例如，结核病和疟疾）。预期结果：我们采用的方法 将开发和分发将提供及时和实用的工具，用于分析多种疾病的数据， 监视系统。对于四种目标感染，我们将回答有关如何提供 特定的监视体系结构和特性-例如系统的可替换空间配置， 具有可变灵敏度和情况确定偏差-可用于改进系统的设计， 实现具体的监测目标。我们将评估监控设计中的关键权衡，并评估 解决特定公共卫生挑战的最佳监测方法。

Project Summary/Abstract Effective surveillance systems are essential to developing targeted, efficient public health responses to infectious disease. In many domestic and global settings, multiple systems are being employed with varying degrees of overlap and with each system designed to achieve different surveillance objectives. The objective of this research project is to develop and distribute a spatio-temporal data integration toolset (“OPTI-SURVEIL”) for analyzing data from multiple surveillance systems, reducing spatial uncertainty by linking data across systems, space and time. With the support and guidance of the US CDC and China CDC (see letters of support), we will develop practical, multi-system analytical tools that will: (1) provide improved estimates of disease burden across space and time; (2) assess individual-level and system-level sensitivity and bias in case ascertainment; and (3) identify redundancies and gaps in current surveillance strategies. Approach: In Aim 1, we will develop spatial- temporal capture-recapture methods for performing data integration at the individual-level. The methods will account for individual-level heterogeneity in case ascertainment and key system-level parameters (sensitivity, bias, dependency). The methods will be extended to consider multiple diseases simultaneously, in order to address the increasing interest in the burden of co-infections. In Aim 2, we will develop spatial-temporal Bayesian hierarchical models for performing data integration using aggregated surveillance data that commonly arise from public surveillance databases. The models will exploit spatial-temporal dependence in county-level case numbers, account for spatial-temporal missing data due to system availability, and probabilistically incorporate information on ascertainment sensitivity and bias at the individual-level and at the system-level. In Aim 3, we will develop simulation-based methods to perform joint evaluation of multiple surveillance system designs. These methods will be applied to optimize a system’s design to maximize case detection while considering resource constraints, system sensitivity/bias, and the presence of other systems. We will apply OPTIM-SURVEIL to existing data in China, and will focus on four infectious diseases of global importance: tuberculosis (TB), malaria, schistosomiasis and hookworm. These diseases exhibit a diverse set of surveillance challenges, including diagnostic accuracy, increasingly rare case counts as elimination is approached, variability in disease severity, and challenges identifying key co-infections (e.g., TB and malaria). Expected Outcomes: The methods that we will develop and distribute will provide timely and practical tools for analyzing data from multiple disease surveillance systems. For the four target infections, we will answer specific questions about how information on specific surveillance architectures and properties—such as alternative spatial configurations of systems that have variable sensitivity and case ascertainment bias—can be used to improve the design of systems for achieving specific surveillance objectives. We will evaluate key tradeoffs in surveillance designs, and assess optimal surveillance approaches for solving particular public health challenges.