Varying coefficient model of longitudinal data of dengue fever in Bandung city

Abstract

Abstract Studies on infectious diseases, especially those transmitted through intermediaries such as dengue fever, are sometimes seen as growth patterns over time. The measurement is repeated over time and hence we have a longitudinal data structure. We are interested in analyzing factors that influencing the dengue fever rate. A Modelling technique is used to determine the effects of covariates, but there are variable where the effects of covariates may also change over time when the effects of covariates change over time. Varying coefficient model is an alternative to model such situation to determine these effects. In general, the model will form a pattern, both linear and non-linear. As the objective function, we propose to use a P-Splines Quantile Objective function such that a flexible dengue pattern can be developed. The proposed technique will be implemented to dengue fever cases in Bandung City. The density of Bandung City is a factor that causes the spread of infectious diseases. In addition, various risk factors for the spread of dengue cases in Bandung are also taken into account, such as rainfall, temperature or temperature and humidity. Another factor that can be a covariate is the lifestyle factor of the people of Bandung, such as clean and healthy living behaviour and healthy house. The result shows existence of a nonlinear pattern, and the fluctuations in the incident value that changes over time. There are changes in the values of the intercept and slope coefficients over time for each level of the quantile.