Covid-19 spatio-temporal prediction using combined graph convolution - LSTM model

Abstract

For the past two years, the world has been going through a tough time because of the global pandemic. The COVID-19 virus spread very quickly throughout the world and dismantled our daily lives, healthcare system, economy, and almost everything. At this point in time, we have lost almost 63 million human lives. From the very beginning of this pandemic, researchers have been trying to predict the nature of the COVID-19 virus and its impact on human life. But there was not sufficient data in the early days to analyze the effect of this deadly virus on a large scale. In spite of that, a lot of research work has been done to predict the future possibilities. Many machine-learning models have been proposed and successfully applied for predictions. Here we collected global data from COVID-19 till date, containing daily new cases, new deaths, cumulative cases, and cumulative deaths country-wise. Since there are several machine-learning algorithms that have been used to predict the COVID-19 virus, in this paper we propose a graph neural network model to make predictions on the number of new cases and deaths. In our model, we will utilize the distance between countries worldwide and try to represent the countries as a graph of structured data. Through our model and obtained predictions, our aim is to provide better knowledge about the spread of the COVID-19 virus and better preparation for any pandemic in the future.