Despite the fact that many countries are gradually removing restrictions in the context of the COVID-19 pandemic, the long-term health of nations continues to depend on tracking the virus and predicting its further spread. Finding the right computer models may not be easy, but two researchers at Binghamton University, New York State University, believe they have an innovative way to solve these problems.
Using data collected from around the world at Johns Hopkins University, Artie Ramesh and Anand Zitharam – both associate professors in the computer science department – have created several forecasting models that take advantage of artificial intelligence.
Machine learning allows algorithms to learn and improve without explicit programming. Models examine trends and patterns from 50 countries where coronavirus infection rates are highest.
“These infections have spread because of measures that were implemented or not implemented, and also because of how some people adhered to the restrictions or not. Different countries of the world have different levels of restrictions and socioeconomic status”.
Anand Zitharam, Associate Professor, Department of Computer Science
For their initial study, Ramesh and Zitharam introduced global infection numbers until April 30, which allowed them to see how their predictions came true before May.
And yet, certain anomalies can lead to difficulties. For example, data from China was not included due to concerns about government transparency regarding COVID-19. In addition, not all countries tracked the spread of the virus; the virus was not a priority.
But there is the main problem – in many countries when calculating the number of infected physicians, they did not record data day by day. If this happens, then there will be a shift in data that the developer model cannot predict.
Scientists fear the second wave of COVID-19: people tired of restrictions will not follow safety recommendations, at least wear masks.
And the developers called the main goal of their research to prepare hospitals and health workers. If they know that in the next three days there will be a surge, and even in their hospitals all the beds are full, then they will need to quickly correct the situation.
As coronavirus spreads around the world, researchers continue to collect data to make their models more accurate. The developers called their work “Ensemble Regression Models for the Short-Term Prediction of Confirmed COVID-19 Cases”.