University of Haifa - Statistics Seminar

Combining Residuals and Control Charts for Detecting Outbreaks in Biosurveillance Data

Inbal Yahav, University of Maryland

June 10, 2009

An important goal of biosurveillance is the early detection of disease outbreaks. Advances in technology have allowed the collection, transfer, and storage of pre-diagnostic information in addition to traditional diagnostic data. Such data carry the potential of an earlier outbreak signature.

Analyzing data for the purpose of detecting outbreaks is composed of two main stages: preprocessing the data to remove explainable patterns, and monitoring the `clean' data (residuals), using control charts, to determine outbreaks. The literature suggests a variety of preprocessing functions and control charts to monitor syndromic surveillance data. However, it is well known that each of these functions is tuned for specific outbreaks. For example, Shewhart charts are optimal for detecting spikes in data; EWMA are better detectors for exponential signals.

When considering syndromic surveillance data streams, the shape of an outbreak signature in usually unknown. Hence it is unclear which function is optimal, and whether there is one function that outperforms all the others. Motivated by these questions, we propose methods to use a combination of functions to analyze and monitor syndromic surveillance data streams. We consider combination methods at two different stages of the monitoring process: combining residuals and combining control charts. For the latter we present a static method where the weight of each chart is predefined. For residual combination we propose an adaptive method based on recent performance.

Back to Seminar Schedule