Qu Z, Sha Y, Xu Q, et al. Forecasting New COVID-19 Cases and Deaths Based on an Intelligent Point and Interval System Coupled With Environmental Variables[J]. Artificial Intelligence-Based Forecasting and Analytic Techniques for Environment and Economics
Abstract: The outbreak of Coronavirus disease 2019 (COVID-19) has become a global public health event. Effective forecasting of COVID-19 outbreak trends is still a complex and challenging issue due to the significant fluctuations and non-stationarity inherent in new COVID-19 cases and deaths. Most previous studies mainly focused on univariate prediction and ignored the uncertainty prediction of COVID-19 pandemic trends, which may lead to insufficient results. Therefore, this study utilized a novel intelligent point and interval multivariate forecasting system that consists of a distribution function analysis module, an intelligent point prediction module, and an interval forecasting module. Aimed at the characteristics of the COVID-19 series, eight hybrid models composed of various distribution functions (DFs) and optimization algorithms were effectively designed in the analysis module to determine the exact distribution of the COVID-19 series. Then, the point prediction module presents a hybrid multivariate model with environmental variables. Finally, interval forecasting was calculated based on DFs and point prediction results to obtain uncertainty information for decision-making. The new cases and new deaths of COVID-19 were collected from three highly-affected countries to conduct an empirical study. Empirical results demonstrated that the proposed system achieved better prediction results than other comparable models and enables the informative and practical quantification of future COVID-19 pandemic trends, which offers more constructive suggestions for governmental administrators and the general public. Copyright © 2022 Qu, Sha, Xu and Li.
Keywords: COVID-19; point forecasting; interval forecasting; artificial intelligence; environmental variables
基金资助：This work was supported by the National Natural Science Foundation of China (Grant No.72004086).
