• time-series forecasting
• RNN
• epidemic
• Deep Learning
• COVID-19
• CNN

In an increasingly globalized modern society, where the life expectancy is keeping improving, humanity have to face an increasing threat: viruses. Viruses have always been part of the human life, however the evolution of the society towards a more globalized one has meant that the spread of these pathogens have become quicker.

Many of such human pathogens have a mammalian and avian origin, when such transmission happens there is the so called spillover. Wild animals live together their own viruses from thousands of years and with evolution have developed an appropriate immune system to face them. The climatic change and the destruction of natural habitats increases the contact between wildlife and humans. This way, the spillover probability increases, exposing humans to unknown viruses with potentially catastrophic outcomes. Recent examples of such events are the filovirus Ebola (discovered for the first time in 1976 in Zaire), the coronavirus SARS (2003) and MERS (2012) and the bird flu.

In recent years the rapid evolution of machine learning technologies have brought remarkable improvements in many fields. In particular, the time-series forecasting is an hot topic which covers many possible applications such as weather, houses and stock prices prediction.

Today, the world is facing a new enemy: COVID19. Actually, there's not any knowledge about it's origin and how to defeat it. Researchers allover the world are trying to find a cure and a vaccine in order to eradicate it. However, being able to predict an epidemic diffusion is of strategic importance. Know in advance how many cases there might be in a given region could allow to anticipate it by reducing the spread and being able to provide better treatments to the infected.