An Early Warning System for Detecting H1N1 Disease Outbreak - A Spatio-temporal Approach

Abstract

The outbreaks of new and emerging infectious diseases in recent decades have caused widespread social and economic disruptions in the global economy. Various guidelines for pandemic influenza planning are based upon traditional infection control, best practice and evidence. This article describes the development of an early warning system for detecting disease outbreaks in the urban setting of Hong Kong, using 216 confirmed cases of H1N1 influenza from 1 May 2009 to 20 June 2009. The prediction model uses two variables – daily influenza cases and population numbers – as input to the spatio-temporal and stochastic SEIR model to forecast impending disease cases. The fairly encouraging forecast accuracy metrics for the 1- and 2-day advance prediction suggest that the number of impending cases could be estimated with some degree of certainty. Much like a weather forecast system, the procedure combines technical and scientific skills using empirical data but the interpretation requires experience and intuitive reasoning.