JMIR Publications

ABSTRACT

Background:

The novel coronavirus was subsequently identified and named as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and the disease caused by SARS-CoV-2 was named coronavirus disease (COVID-19) by the World Health Organization. Control of COVID-19 epidemic has become the global crucial issue. There are abundant research findings on COVID-19-related artificial intelligence (AI) studies to identify and manage potential contacts, but they are either region specific or single-country centered study.

Objective:

COVID-19 pandemic AI system (CPAIS) is a smart system, which automatically comprised databases that contain worldwide COVID-19-related data and each country’s governmental responses toward COVID-19 pandemic, to analyze the dynamic trend of the COVID-19 pandemic, facilitate forecasting and predictive modeling, and produce heatmap visualization of policy measures in different countries.

Methods:

CPAIS integrated the datasets of Oxford COVID-19 Government Response Tracker from the Blavatnik School of Government, Oxford University, and those retrieved from the COVID-19 Data Repository established by Johns Hopkins University Center for Systems Science and Engineering. Four time series models were considered for this study. Regarding entire records in 2020, records of the last 14 days served as the validation set, whereas earlier records served as the training set.

Results:

A total of 171 countries that featured in both the databases were included in the system. CPAIS was developed to explore variations, trends, and forecasts related to the COVID-19 pandemic across several counties. For instance, the number of confirmed monthly cases in the United States reached a local peak in July and another peak of 6,368,591 in December 2020. The dynamic heatmap with policy measures depicts changes in COVID-19 measures for each country. Nineteen measures were embedded within the three sections presented on the website, and only 4 of the 19 measures were continuous measures related to financial support or investment. Deep learning models were used to enable COVID-19 forecasting, and the performance of ARIMA, FNN, and MLP were not stable because their forecast accuracy was only better than LSTM for few countries. LSTM demonstrated the best forecast accuracy for Canada as the RMSE, MAE and MAPE are LSTM are 2272.551, 1501.248 and 0.2723075. ARIMA and FNN demonstrated better performance for South Korea (RMSE = 317.53169 and 181.29894, MAPE = 0.4641688 and 0.2708482, respectively).

Conclusions:

PAIS collects and summarizes information about the COVID-19 pandemic and offers data visualization and deep learning-based prediction. It is a useful and consequential reference resource for any serious outbreak or epidemic that may occur in the future. Also, the two-week machine learning forecasts may serve as a warning sign and highlight current trends in the epidemic that have been made apparent by AI techniques. Moreover, information about the vaccine is also stored in our system.