PSO-Based SEIQRD Modeling and Forecasting of COVID-19 Spread in Italy

Abstract

On 12 March 2020, Coronavirus disease (COVID-19) was declared as a pandemic by the World Health Organization. Efficient disease prevention is challenged by factors, such as population growth, insufficient documentation of clinical effects, dissemination mechanism, and lack of a reliable vaccine. Italy was heavily affected by COVID-19, causing it to have one of the highest COVID-19 deaths in the world. Current statistics show that Italy is on its own way to recover from the second wave of the virus. Therefore, this paper presents a susceptible-exposed-infected-quarantined-recovered-dead (SEIQRD) model to model and predict the spread of the disease in Italy in the near future. The prediction is based on formulating the SEIQRD model using 365-day statistics of the disease. The parameters of the model are estimated using particle swarm optimization (PSO) algorithm. The modeling results agree well with the reported data, with normalized mean absolute error (NMAE) values ranging from 0.058 to 0.214 (average NMAE = 0.115), and normalized root mean absolute error (NRMSE) values ranging from 0.104 to 0.305 (average NRMSE = 0.190). The estimated PSO-based model is then used to predict the future of the disease in Italy in the coming 35 days. The forecasted results indicate that the number of infections will keep reducing, ending the second wave of COVID-19. Such results can assist governments around the world in the process of planning for countermeasures that help reduce the spread of the disease based on forecasted numbers.