A Retrospective Long-Term Analysis of COVID-19 Cases Trends and a Predictive ARIMA Model Through Parameters Screening

Özet

Pandemics are always a source of serious concerns due to the devastating consequences to the communities and nations. Attempts to understand and predict the behavior of outbreaks are challenging as they are largely unpredictable. This research article presents an analysis of weekly COVID-19 cases data for selected Eastern Mediterranean Region (EMRO) country Egypt from January 2020 to May 2024. The study identifies a robust and parsimonious seasonal autoregressive integrated moving average (SARIMA) model for forecasting future trends based on performing comprehensive screening and a comparative analysis of various models. The data reveals a progression of the pandemic in Egypt through multiple waves of varying intensity. Among the models tested, the SARIMA((3,1,0), (0,0,0)) model was identified as the most suitable, demonstrating a strong balance between model fit and parsimony. The model passed all key diagnostic checks, including the Ljung-Box test for residual autocorrelation, with a high p-value of 0.977 at lag 12. This model provides a statistically sound and reliable framework for understanding and predicting the dynamics of the pandemic in Egypt based on the provided dataset. The model’s strength lies in its simplicity and effectiveness, making it a powerful tool for policymakers. Second, the study demonstrates the applicability of the Box-Jenkins methodology to real-world epidemiological data, providing a practical example for similar future studies. The comprehensive screening and comparative analysis of multiple models ensure that the chosen model is not merely a good fit, but the best-fitting and most parsimonious option among the candidates. Finally, the analysis underscores the importance of accurate and consistent data reporting for effective pandemic management and modeling.