Going by the numbers : Learning and modeling COVID-19 disease dynamics

Area of Science:

• Epidemiology
• Computational Biology
• Public Health

Background:

• The COVID-19 pandemic, caused by SARS-CoV-2, has led to significant global health challenges.
• Mitigation measures like lockdowns and social distancing were implemented worldwide to control disease spread.
• Understanding disease dynamics is crucial for effective pandemic management.

Purpose of the Study:

• To develop and evaluate a Long Short-Term Memory (LSTM) based model for predicting COVID-19 trends.
• To provide country and county-level predictions of infection and death rates.
• To quantitatively compare the impact of various mitigation measures on disease dynamics.

Main Methods:

• Utilized a Long Short-Term Memory (LSTM) neural network architecture.
• Trained the model on over four months of cumulative COVID-19 case and death data.
• Performed quantitative analysis of mitigation measure effectiveness using model parameters.

Main Results:

• The LSTM model demonstrated the ability to predict COVID-19 trends at both country and county levels.
• A quantitative comparison of mitigation strategies was performed, highlighting their differential impacts.
• The model's analyses provided insights into infection and death rate trends.

Conclusions:

• The proposed LSTM model offers valuable insights into COVID-19 dynamics.
• The findings can assist policymakers in developing effective mitigation and reopening strategies.
• The model contributes to addressing the global concern of the COVID-19 pandemic.