Space-time Biosurveillance: Sampling, Detection, and Parsimony?

• Activity Number: 436
• Type: Topic Contributed
• Date/Time: Thursday, August 7, 2003 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Andrew Booth Lawson*+
• Companies: University of South Carolina
• Keywords: surveillance ; spatio-temporal ; syndromic ; WSARE ; data mining ; disease

Abstract

The current interest in biosurveillance has led to a variety of proposals for the assessment of changes in the disease incidence environment. Among these are the use of generalized linear mixed models (Lazarus, Kleinman et al. 2002), and the application of spatio-temporal test statistic evaluation (e.g., Kulldorff 2001, Rogerson 2001). However, in general biosurveillance, often a wide range of diseases is to be included, and for early detection there may be a need to break down the detection into subgroups of the population and to monitor ancillary variables (such as pharmaceutical sales and school or job absenteeism). WSARE (what is strange about recent events?) (Wong, Moore et al. 2002) is a data mining algorithm which uses a variety of rule-based and statistical procedures to mine extensive health record databases. Currently, no attempt is made to mine spatial information. In this presentation, a review of some of the issues at the interface between data mining syndromic surveillance and spatial epidemiology will be given. The importance of using flexible general models for multi-focus surveillance of maps will be stressed (Lawson 2003).