Project Example: Outbreak detection of Campylobacteriosis time series

Research Objective

In order to detect dispersed or nationwide infectious disease outbreaks, public health institutions are monitoring reported infectious disease reports. In this context, fast and efficient statistical methods are essential as components of the data-mining process in detecting aberrations in routinely collected surveillance data. Those data exhibit time trends, strong seasonality and various artefacts. This project was concerned with the development of an automated algorithm that analyses time series data of infectious disease reports. Based on all past observations, the seamless procedure is computes a threshold to decide whether the observation of the current disease counts is unusaul. The developments were motivated by the routine monitoring of Campylobacteriosis in Germany, the leading cause of enteritic illness in industrial countries. Since foodborne disease spreading is strongly influenced by weather conditions, the weekly mean absolute humidity is considered as potential covariate to correct for epidemiological background information.

Statistical Methodology


Open source implementation using integrated nested Laplace approximation (INLA) surveillance::boda() is available on R-forge.

Related Publications