A time-varying state-space model approach for Covid-19 infection in Malta

Abstract

The coronavirus pandemic has quickly escalated into a global public health emergency. The rapid spread of SARS-CoV-2 has emphasised the need of developing a number of mitigation strategies under conditions of extreme uncertainty. While several recent works have provided projections of the disease, most utilise time-independent parameters which may fail to capture the dynamic transmission and removal processes that are governed by containment protocols implemented at various stages of the pandemic. This dissertation makes use of an extended susceptible-infectious-removed state-space (eSIR) model which incorporates a time-varying transmission rate to estimate a time-dependent disease reproduction number, which may reflect the efficacy of infection control strategies. Time-dependent modifiers for the transmission rate are estimated via a time series SIR model. We apply these methods to analyse and forecast the daily COVID-19 infection prevalence and probability of removal in Malta using Bayesian framework through MCMC. We further obtain estimates of key epidemiological parameters, including the rate of transmission and removal and the expected total incidence by the end of the forecasting period. Our findings suggest that enforcement of preventive and control policies may result in a substantial decrease in infection and reproduction number such that the disease in Malta can eventually die out in the near future.