A Computational Testbed for Biosurveillance Methods

Stanford University, Stanford CA

Investigators

Linked publications & trials

Abstract

In the last decade, public health specialists have become increasingly concerned with the possibility of[unreadable] outbreaks resulting from bioterrorism and emerging infectious diseases. With the intent of enabling rapid[unreadable] detection of incipient epidemics, many groups have implemented surveillance systems that operate on a[unreadable] variety of "prediagnostic" data sources and that employ different kinds of outbreak detection algorithms.[unreadable] Unfortunately, evaluation of these systems, including their data sources and detection algorithms, has not[unreadable] accompanied the flurry of implementation activity. A major reason for the lack of evaluation is the difficulty[unreadable] encountered in configuring surveillance systems to analyze novel data sources and to employ novel[unreadable] detection algorithms without the need for endless reprogramming. We will develop an ontology-driven[unreadable] computational test bed that will enable the critical and easy evaluation of these surveillance data sources[unreadable] and analytic methods. To meet this goal, we propose the following four specific aims: (1) Develop a[unreadable] classification of analytic methods and surveillance data used in outbreak detection. (2) Build a software[unreadable] control structure for executing evaluation studies of outbreak detection methods. (3) Encode software[unreadable] methods for outbreak detection and evaluation. (4) Conduct specific evaluation studies with data from[unreadable] Boston and Montreal healthcare organizations using the test bed. The underlying methodologies that we will[unreadable] develop for integrating surveillance data streams and for allowing experimentation with analytic methods will[unreadable] enable the development of a new generation of surveillance systems that will take full advantage of the[unreadable] expanding public health information infrastructure. The computational architecture that we propose will serve[unreadable] as a model for the next generation of such systems.[unreadable]