Forecasting the Spread of the Sixth Wave of COVID-19 Epidemic in Southern Iran: An Application of ARIMA Models

Abstract

Background: Through the fifth wave of the Covid-19 outbreak in Jahrom, the fatality and incidence of the virus increased. The quick spread of infection is one of the causes of this dreadful situation. Therefore, recognizing the future epidemic trend can be a useful instrument to decrease mortality and morbidity. This study aimed to determine the time trends and select the best model to predict the sixth wave of the COVID-19 outbreak using ARIMA models.Methods: We used daily data of 9533 hospital cases (Suspected and PCR-confirmed COVID-19 cases) between 4th March 2020 and 31st December 2021. Nine different ARIMA models were fitted to our data. Autocorrelation functions (ACF) and partial autocorrelation (PACF) plots were used to determine model parameters. Likelihood-ratio test for comparison of the reduced and full model was used. In addition, Akaike Information Criteria (AIC) was also used to choose the final model. Data were analyzed by STATA 14 software with a significant level of 0.05.Results: The ARIMA (3, 0, 3) model was selected among the potential models, with lower AIC (999) and MAPE (3.18%) values. This model showed that the daily number of hospitalized patients may increase from 5.85 (2.16-15.79) to 8.55 (1.47-49.48) in two months. By March 01, 2022, the predictable daily hospitalized cases could reach 468.36 (03.79-2209.88).Conclusion: Time series models is a useful tool for predictingthe hospitals’ admission trend during an epidemic. Thus, they can be used as early warning models in the readiness of hospital systems during epidemics.

Keywords

• covid-19
• predicting
• sixth wave
• spread
• time series modeling