Varicella incidence rate forecasting in Bogotá D.C. (Colombia) by stochastic time series analysis

Abstract

This work presents a specific implementation of the Allard approach to the epidemiological time series analysis by ARIMA and SARIMA modeling, intended to describe and predict the epidemic behavior of Varicella in the city of Bogotá D.C. (Colombia). Model selection and preliminary forecast evaluation supported on the official accounts of Varicella incidence rate are performed and reported for the interval 2010–2015. This approach yields a SARMA(3,0,1)(1,0,1) model, whose forecasting results were evaluated against real data of the year 2016. Performance comparison with alternative models and their potential use in the support of epidemic surveillance are also discussed.