A Deep Learning Method to Forecast COVID-19 Outbreak.

Area of Science:

• Epidemiology
• Computational Science
• Machine Learning

Background:

• The COVID-19 pandemic, emerging in late 2019, has profoundly disrupted global lifestyles.
• Accurately predicting pandemic trajectories and end-points remains challenging due to evolving parameters.

Purpose of the Study:

• To develop and compare prediction models for pandemic behavior using machine learning techniques.
• To evaluate the efficacy of Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) networks for pandemic simulation.

Main Methods:

• Support Vector Regression (SVR) was employed, utilizing a function to estimate input-to-real number mapping based on a training model.
• Long Short-Term Memory (LSTM) networks, a type of Recurrent Neural Network (RNN), were utilized for their ability to learn long-term dependencies.
• Both SVR and LSTM techniques were applied to simulate the behavior of the pandemic.

Main Results:

• Simulation results indicated that LSTM models provided more realistic pandemic behavior predictions.
• The study specifically highlights the effectiveness of LSTM in the Indian scenario.

Conclusions:

• LSTM networks offer a promising approach for accurate pandemic prediction due to their capacity for learning long-term dependencies.
• The findings suggest LSTM as a valuable tool for understanding and forecasting pandemic dynamics, especially in specific regional contexts like India.