Artificial intelligence in pursuit of SARS virus host species

Artificial intelligence can predict which viruses will infect humans in the near future, according to an international study.

An international team of researchers led by scientists from Georgetown University has shown that artificial intelligence can predict with a excellent probability which viruses are going to infect humans in the near future and where on Earth they are most likely to appear and spread.

The main hosts of viruses of type SARS identified

“If you want to find these viruses, you have to start by profiling their hosts – ecology and evolution,” said Colin Carlson, lead author of the study. His team spent 18 months validating several predictive models to determine which bat species could be hosts for SARS-CoV-2 -like viruses – who are at the origin of the current Covid-19 pandemic.

“With artificial intelligence, we can turn data on bats into concrete predictions, i.e. knowing where to look for the next case of SARS “, He declared elsewhere.

“One of the most important results of our study is a selective list, based on data, species of bats that should be the subject of extensive monitoring, ”said Daniel Becker, one of the co-authors of the paper.

“After identifying these likely hosts, the next step is to invest in a monitoring program to understand where and when the betacoronaviruses (the group to which the viruses similar to SARS belong) are likely to appear, ”he said elsewhere.

Eight predictive models based on artificial intelligence

During the first quarter of 2020, the research team trained eight different statistical models to to predict which animal species were likely to harbor betacoronaviruses.

For over a year she then stalked 40 species of bats in order to validate the initial predictions and update the models dynamically.

The researchers found that artificial intelligence models using data on the ecology and evolution of bats were particularly efficient to predict new hosts. On the other hand, those based on sophisticated mathematics, but which rely less on biological data, do not convince, giving more or less random results.

For Greg Albery, also a co-author, the COVID-19 epidemic played a major role in the success of this interdisciplinary study combining artificial intelligence, advanced mathematics and animal biology. “Outside of a pandemic, we would never have learned so much about these viruses in such a short time. A decade of research has been condensed in about a year of publication, which means we’re actually able to show that these predictive AI models work ”.


Share on facebook
Share on twitter
Share on linkedin

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

On Key

Related Posts


#HotwireHotelGoals #TakisTransformation #movie #SimlishSessions #fy