Collaborative Research for Developing ATD: Bayesian Methods in Syndromic Surveillance: CAR Models and Computational Implementation

该合作团队开发并实施了以早期发现疾病暴发为目标的综合征监测的贝叶斯公式计算。该研究涵盖了统计理论和方法，以及算法和计算科学的问题。条件自回归模型或CAR模型构成了该公式的基础，以适应空间和时间效应。主要的挑战是可扩展性（7500家美国主要医院连接到报告网络中）、多变量报告、数据中复杂的依赖结构、低数据质量以及纳入区域和其他协变量。统一的主题是不确定性的原则性计算。本研究以具有国家意义的应用为框架。疾病的早期发现带来了巨大的社会和经济效益。风险是巨大的：美国技术评估办公室估计，如果不迅速发现，100公斤炭疽孢子在华盛顿特区逆风释放，将导致多达300万人死亡和1万亿美元的经济损失。这项研究产生了可操作的、易于解释的信息，为决策者提供了支持：一种特定疾病出现在美国某地或某城市的可能性的明确陈述，以及相关的不确定性。误报率得到控制，使其与信号后调查的可用资源相称，并考虑到误报的心理和财务影响等后果。

The collaborative team develops and implements computationally of a Bayesianformulation of syndromic surveillance targeted at early detection of diseaseoutbreaks. The research spans statistical theory and methodology, as well asissues of algorithms and computational science. Conditional autoregressivemodels, or CAR models, form the basis of the formulation, in order toaccommodate spatial as well as temporal effects. The principal challenges arescalability (7,500 major U.S. hospitals are linked into the reportingnetwork), multivariate reports, complex dependence structure in the data, lowdata quality, and incorporation of regional and other covariates. The unifyingtheme is principled calculation of uncertainties.The research is framed by an application of national significance. There isenormous social and economic benefit from early discovery of disease. Thestakes are immense: the US Office of Technology Assessment estimates that arelease of 100 KG of anthrax spores upwind of Washington, DC would, if notdetected rapidly, lead to as many as three million deaths and a trilliondollars of economic losses. The research produces actionable, easilyinterpretable information that supports decision makers: a clear statement ofthe probability that a particular disease is present somewhere in the US, orpresent in a particular city, together with the associated uncertainty.False alarm rates are controlled, making them commensurate with resourcesavailable for post-signal investigation and taking into accountconsequences such as the psychological and financial impacts of a false alarm.