AI helps clear up COVID-19 data
As governors begin to make decisions about reopening the economy, Americans are left to wonder whether they should follow their state government’s lead – or make their own decisions about when to return to normal during this overload of COVID-19 data, but AI can help.
One problem for the average person: How to decipher the multitudes of data about COVID-19 and evaluate whether the country or any particular state is – or is not – flattening the curve.
“It’s easy to find tons of data online with charts and graphs, but all those numbers can be overwhelming,” said Sharon Daniels, Chief Executive Officer of Arria, which specialises in a form of artificial intelligence known as Natural Language Generation (NLG). “You see a line on a graph, but what is it telling you?”
Daniels’ company is among those trying to simplify that complex chore for Americans, using artificial intelligence to transform that raw data into an easy-to-understand narrative. To this end, Arria is involved with two online initiatives – the COVID-19 Live Report and the COVID-19 U.S. Tracking Report – that give Americans access to NLG as they try to grasp all the information coming their way from scientists, government officials, and the media.
Each of these free dashboards allows anyone – from government leaders to journalists to citizens – to review up-to-date COVID-19 data, along with critical insights transformed into writing by Arria’s Natural Language Generation software. The software uses language analytics and computational linguistics to ‘think’ like a writer, pulling the most important information to the top of the narrative, providing critical insights, and giving meaning to the tabulated reports and visualisations.
Just as an example, a resident of Pulaski County, KY, who checked in on April 23rd would have learned that in their community the previous day ‘there were two new cases and no deaths reported. During the past seven days, cases have increased by seven, which means the seven-day rolling average for cases is one.’
No human wrote those sentences. They were penned automatically by the NLG software.
As Arria and others do their part to help Americans work their way through the sea of information, there is evidence that such assistance is both needed and wanted:
A Gallup poll shows lots of confusion about the state of the virus in the US, with Americans reaching no consensus on how they think things now stand; 41% say the situation is getting better, 39% say it is getting worse, and 20% say it is staying the same.
A 2017 study of the US public’s understanding of the 2014 Ebola outbreak in West Africa found that most people are good at assessing risk when information is communicated accurately and effectively. That study also found that Americans want accurate and honest information, even if that information might worry people.