Greece facing Covid
Between June and November 2020, Greece reopened its country to vacationers and tourists, and faces a problem: how best to use its limited PCR testing capabilities to limit the risk of Covid-19 contamination?
As of August 2020, it had welcomed more than 1.6 million international passengers at its airports, according to Tourmag. It would have been impossible to test everyone. The authorities decided to implement a machine learning system, a more accurate reinforcement learning, so that passengers are tested.
Greece and Covid 19: AI as a backup!
This type of model is trained to improve its decision as it goes along by capitalizing on the accumulated results of its previous decisions. An article published on September 10 in the scientific journal Nature provides an update on the results of this experiment.
Three variables: country, gender and age
Called Eva, the system has been used in 40 airports, ports and land border crossings, with the support of 400 health personnel to perform the tests.
To determine who to test first, it groups travelers into categories based on their origin, gender, and age, and compares them to available prevalence statistics on these different groups based on tests already conducted. Some data, such as travelers’ occupations, were deliberately excluded from the development of the algorithm due to concerns about personal data.