Disease Surveillance in Real Time: Geotemporal Methods

描述（由申请方提供）：真实的实时数据收集和分析将大大加强区域和国家层面的公共卫生监测。 虽然症状和诊断数据存储在行政保健数据库中，但目前没有自动化系统来整合这些数据，以便及时发现异常疾病模式，最近的生物恐怖袭击突出了这一问题。 我们的目标是通过与现有的和快速扩展的数据采集基础设施合作来解决这个问题，以开发一旦患者开始出现在医疗保健站点时就尽快识别疾病群所需的分析工具。 我们已经通过真实的实时虚拟整合多个医院急诊科（艾德）数据库，建立了一个监测网络。 这提供了一幅区域人口疾病模式的图景。 此外，在其中一家医院，我们收集了回顾性数据库，其中包含多年的历史数据，这些数据对于建立分析模型的正态性至关重要。 因此，该项目的目标是通过使用随时可用的实时信息来加强监测系统，从而提高这些系统更快发现疾病爆发的能力。 由于迫切需要迅速确定炭疽等生物战剂的受害者，我们将集中精力确定出现呼吸系统综合征的患者群。 拟议的研究将首先建立疾病的正常模式，然后建立模型，使这些模式的偏差检测与最小的误报。 一旦这些方法建立起来，它们就可以应用于任何其他有意义的综合征。

DESCRIPTION (provided by applicant):Real time data collection and analysis would substantially strengthen public health surveillance at both regional and national levels. Although symptom and diagnostic data are stored in administrative health care databases, there currently are no automated systems for integrating this data so that abnormal patterns of disease can be detected in a timely manner, a problem highlighted by recent bioterrorist attacks. We aim to address this by working with an existing and rapidly expanding data acquisition infrastructure to develop the analytic tools needed to recognize disease clusters as soon as possible once patients begin appearing at health care sites. We have already established a surveillance network by virtually integrating multiple hospital emergency department (ED) databases in real time. This provides a picture of regional population patterns of disease. Furthermore, at one of these hospitals, we have assembled retrospective databases with many years of historical data critical for establishing normality for the analytic models. Thus, the goal of this project is to strengthen surveillance systems by using readily available information in realtime, thereby increasing the power of these systems to detect disease outbreaks sooner. Because of the urgent need to quickly identify victims of biowarfare agents such as anthrax, we will concentrate on identifying clusters of patients presenting with respiratory syndromes. The proposed study will first establish normal patterns of disease and then build models that enable the detection of deviations from these patterns with a minimum of false alarms. Once these methods are established, they can be applied to any other meaningful set of syndromes.