Facebook has published a paper giving a forecast of COVID-19 spread across the U.S using AI. The company posted the news claiming that its forecast has seen a strong performance compared to other forecasts.
It is a wonder Facebook has time to enter the area of COVID-19 forecasting with everything the company has going on. With new features rolling out left right and centre as well as the U.S. election, Facebook is going through a busy period.
Most recently the platform brought out a new tabbed news feed for some android users, in order to provide better control over what users see. The company has also faced a number of issues with the spread of misinformation in the run-up to the election. However, it claims that they are much better prepared to deal with the issues that may arise.
An accurate COVID-19 forecast is potentially a very useful tool given the decisions that need to be made regarding hospital capacity and lockdown restrictions. If Facebook has a successful one using AI then this forecast could have an impact on the COVID-19 pandemic.
Facebook publishes AI-driven COVID-19 forecast
Facebook’s system aims to predict infection spread 14 days in advance so it is not a long-term predictor. The main areas it focuses on are the nature of the disease and the social factors that influence its reach.
The company has also crafted a “neural autoregressive model” which aims to separate regional and disease-specific elements of the data. This means that a spike in one country could affect neighboring counties which this research accounts for.
Researches involved believe this model could be more effective than others out there. This is because it offers a more complex model but also makes few assumptions than other predictions out there.
Right now the forecast only makes predictions across the U.S. However, the company hopes to translate the model across to Europe to offer something similar. They hope to do this in conjunction with Barcelona’s Universitat Politècnica de Catalunya.
It is still not totally clear as to whether this prediction will be accurate or not. Its success depends on this quite heavily as well as whether decision-makers adopt it.
However, if it does stand up to the rigorous of academic scrutiny as well as accuracy and people start using it then Facebook could be onto a winner here. This forecast does genuinely have the possibility of making a difference to the pandemic at large.