Bayesian spatial conditional autoregressive (CAR) Leroux model of Covid-19 cases in Makassar, Indonesia

Abstract

The number of positive patients for the Coronavirus disease-2019 (Covid-19) is growing exponentially. South Sulawesi Province, one of the provinces in Indonesia, has the highest number of Covid-19 cases outside Java Island. Makassar City as the provincial capital of South Sulawesi has the highest number of positive confirmed Covid-19 cases in South Sulawesi. This study aims to estimate the relative risk of Covid-19 cases in Makassar by comparing several Bayesian Spatial Conditional Autoregressive (CAR) Leroux models. Data on the number of confirmed positive cases of Covid-19 from March 20, 2020, to August 30, 2021, in every sub-district (15 sub-districts) in Makassar City are used. In addition, data on the number of populations from each sub-district are also used to calculate the expected value of the occurrence of Covid-19 cases. The selection of the best model is based on several criteria, namely Deviance Information Criteria (DIC), Watanabe Akaike Information Criteria (WAIC), and residuals from Moran’s I Modification (MMI). The Bayesian spatial CAR Leroux model with hyperprior Inverse-Gamma (0.5;0.05) is preferred to model the confirmed Covid-19 cases in Makassar city. Ujung Pandang has the highest relative risk of Covid-19 while Sangkarrang Island has the lowest relative risk of Covid-19. These results may help policy makers in decision-making.