Predicting the incidence of human campylobacteriosis in Finland with time series analysis

Abstract

Human campylobacteriosis is a common bacterial cause of gastrointestinal infections. In this study, we tested whether spectral analysis based on the maximum entropy method (MEM) is useful in predicting the incidence of campylobacteriosis in five provinces in Finland, which has been accumulating good quality incidence data under the surveillance program for water- and food-borne infections. On the basis of the spectral analysis, we identified the periodic modes explaining the underlying variations of the incidence data in the years 2000-2005. The optimum least squares fitting (LSF) curve calculated by using the periodic modes reproduced the underlying variation of the incidence data. We extrapolated the LSF curve to the years 2006 and 2007 and predicted the incidence of campylobacteriosis. Our study suggests that MEM spectral analysis allows us to model temporal variations of the disease incidence with multiple periodic modes much more effectively than using the Fourier model, which has been previously used for modeling seasonally varying incidence data.