Biosurveillance using control charts and scan statistics
Επιδημιολογική επίβλεψη με χρήση διαγραμμάτων ελέγχου και στατιστικών σάρωσης
In the present study, is described a monitoring system which is developed to identify unusually large increases in time series of infectious disease counts (outbreak detection) compared to the expected number of cases. The designed two-phase monitoring system consists of (i) successful integration of count time series following generalized linear models, in order to provide dynamic forecasting of future expected disease counts and (ii) using SPC methods for tracking the forecast errors from the fitted models. Analysis for this study was illustrated on time-dependent count data which reflect the total number of infectious diseases from January 2005 through December 2012 in different geographical areas in Greece and were reported weekly to the Hellenic Center for Disease Control and Prevention (HCDCP). The systematic methodology that is developed, is capable of detecting aberrations in infectious disease patterns, facilitating a timely public health response and it can be generalized to other healthcare settings.