B0512
Title: Prediction of COVID-19 incidence in Africa using Bayesian a hierarchical smooth transition autoregressive model
Authors: Geoffrey Singini - University of Malawi (Malawi)
Samuel Manda - University of Pretoria (South Africa) [presenting]
Abstract: Disease incidence forecasting informs disease control policies, resource allocation and preparedness level of the health care system. A common approach for forecasting infectious disease incidences is based on linear time series models such as the autoregressive moving average (ARMA)-type models. Due to the time dynamics of infectious diseases, these linear time series are limited in scope. Using a simulation study, we show the performance of nonlinear smooth transition autoregressive (STAR) models in capturing the nonlinear dynamics in infectious disease data in comparison to linear time series models. The capabilities of STAR-type models are demonstrated with an application to forecasting COVID-19 incidence in African countries, with the country-specific nonlinear dynamics captured by a logistic transition function. Both in-sample and out-sample COVID-19 incidence predictions are used. The parameters of the resulting model are estimated using the Bayesian hierarchical modelling approach.