Mapping sub-districts-level and predicting spatial spread of COVID-19 death case in Jakarta

Abstract

Currently, the number of deaths caused by COVID-19 continues to increase significantly, especially in areas with high population density and mobility such as Jakarta, Indonesia. Spread of infectious diseases has a spatial closeness which results in cases of COVID-19 deaths also have a spatial dependency which is influenced by cases of deaths in the surrounding area. This study aims to map and predict the number of COVID-19 deaths using the Bayes Linear Mixed Model (LMM) method involving spatial random effects. The response variable is number of deaths and the explanatory variable are number of positive cases of COVID-19 and population density with sub-district area units in Jakarta. Response variable is divided into 6 schemes (PSBB 1, PSBB Transition 1, PSBB 2, PSBB Transition 2, PPKM 1 and PPKM 2) which is adjusted to the policies and social distancing activities from Jakarta provincial government, and assumed to have a normal distribution with INLA (Integrated Nested Laplace Approximation) inference approach. Some important results from this study are: in all 6 social distancing schemes, the number of positive cases of Covid-19 has a significant effect on the increase in number of deaths, while population density has a significant effect along with the increasing variance value of response data. The Bayes LMM has successfully mapped the spread of COVID-19 cases with the best RMSE value of 3.31. The mapping results show that several sub-districts with high population density and sub-districts located on Jakarta border have a high risk of death. Furthermore, the PSBB and PSBB Transition social distancing schemes are considered to be quite effective in suppressing the diversity number of deaths. However, it is different from the PPKM scheme where it is predicted that there will be an increase in the number of high-risk districts for COVID-19 up to 51% per day.