Forecasting of dengue hemorrhages fever cases with autoregression distributed lag model using Poisson regression approach

Abstract

Dengue hemorrhages fever (DHF) is an infectious disease epidemic in the community. DHF is caused by a virus that is transmitted through the bite of mosquitoes Aedes sp. The aim of this research was to develop and validate a forecasting model that could predict dengue fever (DHF) cases in Bandung, West Java, Indonesia. Model of dengue fever is developed using time series Poisson multivariate regression. Poisson regression is a generalized linear model form of regression analysis used to model count data and contingency tables. Poisson regression assumes the response variable has a Poisson distribution and the logarithm of its expected value can be modeled by a linear combination of unknown parameters. Here, a time series Poisson multivariate regression model was developed using monthly mean temperature and cumulative rainfall over the period 2001–2016. Cross-correlation function between response and predictor is used to decide the lag parameter optimum in autoregression distributed lag model. Model selection and validation were based on the Akaike’s information criterion, mean square error, mean absolute percentage error, and residuals diagnostic test. The R-square obtained is 0.78, means that the temperature, rainfall, past dengue fever cases, season, and trend could explain 78% of the variance of monthly dengue distribution.