Long-Term Prediction of Large-Scale and Sporadic COVID-19 Epidemics Induced by the Original Strain in China Based On the Improved Non-Autonomous Delayed SIRD and SIR Models

Abstract

The COVID-19 virus emerged suddenly in early 2020 and spread rapidly, causing a significant impact on national health. To achieve our goal, we introduce a time-delay factor based on the traditional SIR/SIRD model and perform a sliding average on the data collected from the official website during the data preprocessing stage. The results of this study are in very good agreement with the actual evolution of COVID-19, and the prediction accuracy can all be controlled within 3%. From our model parameter perspective, under strict isolation policies, the transmission rate of COVID-19 in China is relatively low and still significantly reduced, indicating that government intervention has had a positive effect on epidemic prevention in the country. Besides, our model is also successfully applied to predict the outbreaks caused by the SARS virus in 2003 and the COVID-19 outbreak induced by the Omicron virus in 2022, demonstrating its wide application and effectiveness. This work facilitates timely measures and adjustment of medical resources in various regions, ultimately helping to reduce economic and social losses.