Monitoring COVID-19 pandemic in Saudi Arabia using SEIRD model parameters with MEWMA

Abstract

Background: When the COVID-19 pandemic hit Saudi Arabia, decision-makers were confronted with the difficult task of implementing treatment and disease prevention measures. To make effective decisions, officials must monitor several pandemic attributes simultaneously. Such as spreading rate, which is the number of new cases of a disease compared to existing cases; infection rate refers to how many cases have been reported in the entire population, and the recovery rate, which is how effective treatment is and indicates how many people recover from an illness and the mortality rate is how many deaths there are for every 10,000 people. Methods: Based on a Susceptible, Exposed, Infected, Recovered Death (SEIRD) model, this study presents a method for monitoring changes in the dynamics of a pandemic. This approach uses a Bayesian paradigm for estimating the parameters at each time using a particle Markov chain Monte Carlo (MCMC) method. The MCMC samples are then analyzed using Multivariate Exponentially Weighted Average (MEWMA) profile monitoring technique, which will “signal” if a change in the SEIRD model parameters change. Results: The method is applied to the pre-vaccine COVID-19 data for Saudi Arabia and the MEWMA process shows changes in parameter profiles which correspond to real world events such as government interventions or changes in behaviour. Conclusions: The method presented here is a tool that researchers and policy makers can use to monitor pandemics in a real time manner.

Keywords

• COVID-19 pandemic
• Statistical process monitoring
• Augmented Markov chain Monte Carlo
• Exponentially weighted moving average control
• Multivariate control