A COVID-19 Pandemic Artificial Intelligence-Based System With Deep Learning Forecasting and Automatic Statistical...

Area of Science:

• Epidemiology and Public Health
• Artificial Intelligence in Healthcare
• Data Science and Predictive Modeling

Background:

• The COVID-19 pandemic caused over 79.2 million cases and 1.7 million deaths globally.
• Effective control of the pandemic requires understanding global trends and national policy responses.
• Limited studies have comprehensively analyzed global COVID-19 trends alongside country-specific policy measures.

Purpose of the Study:

• To develop an online artificial intelligence (AI) system for analyzing COVID-19 pandemic dynamics.
• To facilitate forecasting and predictive modeling of the pandemic's trajectory.
• To visualize policy measures implemented by 171 countries using a dynamic heat map.

Main Methods:

• Integrated data from Oxford's COVID-19 Government Response Tracker and Johns Hopkins' COVID-19 Data Repository.
• Employed four forecasting techniques: ARIMA, FNN, MLP, and LSTM.
• Utilized a 1-year time series, with the last 14 days as validation and earlier data for training.

Main Results:

• The COVID-19 Pandemic AI System (CPAIS) was developed, covering 171 countries.
• LSTM demonstrated superior forecasting accuracy for Canada, while ARIMA and FNN performed better for South Korea.
• A dynamic heat map visualized policy measure variations across countries, highlighting 19 measures, with 4 being continuous financial support measures.

Conclusions:

• The CPAIS provides data visualization and deep learning-based predictions for the COVID-19 pandemic.
• The system serves as a valuable reference for predicting serious outbreaks or epidemics.
• Daily updates, including vaccination data, ensure the system reflects the evolving pandemic dynamics.