Teams turn to AI to tackle the spread of Covid

Abstract

This article discusses the XPrize Pandemic Response Challenge 2020. For this challenge participants were tasked with using artificial intelligence (AI) to create models that could analyse data, and then use it to predict the future of local Covid-19 transmission rates, whilst also allowing the potential effect of hypothetical policy decisions to be seen, and recommendations to be made for future policies. A variety of models were submitted including simulation-based epidemiological modelling mixed with classical machine learning, probabilistic mixture models with decision-tree algorithms, and machine learning supporting human expertise. Regular predictions were made about how well different measures would reduce infection rates and impact economies, and would be tracked using real-world data, with each team's models being run through a reality simulator to test their predictive powers further, along with testing a variety of 'what-if' scenarios. The winners were VALENCIA IA4COVID19 team, using a deep neural network and machine learning to make its recommendations, and JSI vs COVID, using a labelled traditional SEIR epidemiological model with machine learning to predict infections.