Global Prediction Model for COVID-19 Pandemic With the Characters of Multiple Peaks and Local Fluctuations

Background: With the global spread of COVID-19, the time-series prediction of COVID-19 outbreaks has become a research hotspot. Unlike previous viruses, the spread of COVID-19 is characterized by a new pattern of long time series, large fluctuation, and multiple peaks. However, Traditional prediction models are limited to epidemic curves with short time series, single-peaked, and smooth symmetric forms. Secondly, most of these models have more unknown parameters, which cannot be supported by the actual data and bring greater ambiguity and uncertainty to the prediction results. Finally, there are still major shortcomings in the integrated description of multiple factors such as human intervention, environmental factors, and transmission mechanisms of infectious disease.Methods: A prediction model with only infected and removed populations was established to better predict the future trend of the COVID-19 pandemic. Firstly, a long time-series process of COVID-19 spread was segmented based on local smoothing. In addition, the change in infection rate at each stage was quantified by using a Logistic growth function that quantitatively described the comprehensive effect of natural and human factors on virus infectivity. Secondly, non-linear variable factors and abnormal value of NO2 concentration were introduced to describe the number of people to avoid infection under human control measures, which could quantify the impact of artificial prevention and control on the temporal evolution of the epidemic.Results: Through experiments and analysis of epidemic data from five countries, the United States, the United Kingdom, India, Brazil, and Russia, the new model can not only better describe the effects of human interventions on COVID-19 from both qualitative and quantitative aspects but also better predict the temporal evolution of COVID-19 dynamics with high correctness and rationality, especially for COVID-19 pandemic with large and local fluctuation.Conclusion: This study proposed a novel infectious disease model to better predict COVID-19 and quantify the impact of human interventions. Simulation results of epidemic data showed that the model could achieve better prediction results, which can provide valuable assistant decision-making information for global epidemic prevention and control.Funding Information: This study was funded by the National Key Research and Development Program of China (2018YFB0505304) and the National Natural Science Foundation of China (Grant No. 41671409). Declaration of Interests: The authors declare that they have no competing interests.