Empirical Bayes Hierarchical Modelling and Mapping of HIV/AIDS

Abstract

Disease mapping of incidence and/or prevalence in epidemiology by statistical modeling play an important role in pointing out the spatial risks on a map and describing the causal relationship between outcomes and the potential risk factors. Major problem of public health in Thailand is HIV/AIDS infection. Empirical Bayes method of hierarchical data was aimed to fit the data an HIV/AIDS mapping and to cope with the incidence model of risk factors by using HIV/AIDS infection data of new diagnosis in Thailand 2013 to 2017 from the National AIDS Program (NAP), collected by the National Health Security Office (NHSO). Under the previously empirical data, prior estimation is fitted well with goodness-of-fit values of the Kolmogorov-Smirnov (KS) test. Empirical Bayes Poisson hierarchical approach performs well in both HIV/AIDS mapping and modeling of incidence among risk factors, such as gender and age group. The best-fitted model was the interaction effects and found that in 2015, 2016 and 2017, HIV/AIDS infection rate is high risk at male aged 24-49 years. The top seven provinces with the highest risk (infection rate $\\gt 7.125$%) were Nakhon Nayok, Samut Prakarn, Chumphon, Pathumthani, Singburi, Phuket, and Buri Ram, respectively.