The COVID-19 pandemic has raged for over two years and is far from over. The discovery of every new strain leads to panic among the global community and increases the possibility of lockdowns and new travel restrictions. However, curbing international travel has detrimental effects on a country’s economy due to decreased revenues from the travel industry, not to mention the countries that depend heavily on the migrant workforce for their industries. A case study on the effects of the pandemic on the Australian economy clearly shows losses can be attributed to the travel and education sectors due to a decrease in the number of tourists and international students. An estimated decrease in revenue of US $1 trillion in the travel industry of European countries, along with a reduction of 19 million jobs, has crippled most tourist economies.
The usual screening protocol followed and its drawbacks
Countries have had difficulty deciding how to best open international borders for travel without further surges in COVID-19 cases. Most countries, other than Greece, adopted colour grouping of nations depending on the population level epidemiological metrics. With travellers from ‘white-listed countries’ exempt from any restrictions, those from ‘grey-listed countries’ having to provide proof of recent negative RT-PCR reports, people from ‘red-listed countries’ had to undergo mandatory quarantine on arrival and travellers of ‘black-listed countries’ were forbidden from entering the country. There are several drawbacks to this grouping process, and better screening methods are necessary to reduce the risk of reopening borders.
How Greece leveraged AI for reopening its borders
Artificial intelligence has been applied to deal with the different challenges faced during the COVID-19 pandemic, from predicting ICU admission of patients to providing verified leads for life-saving supplies. When faced with the problem of restarting non-essential travel in July 2020 to save its tourism-dependent economy, Greece relied on AI for determining the best strategy for this purpose.
It all started with the Greek Prime Minister Kyriakos Mitsotakis accepting the assistance of Kimon Drakopoulos, an Assistant Professor of the Data Sciences and Operations department at USC Marshall School of Business. Specifically, Greek authorities wanted help in reopening borders while maintaining utmost safety.
On the advice of Drakopoulos and his colleagues, Greece authorities used a machine learning algorithm based system to determine which foreign travellers entering the country needed to be tested for coronavirus infection. The machine learning system is known as Eva, which according to researchers, was able to detect more than two to four times the number of infected travels than random testing.
International travellers coming to Greece had to fill up a form 24 hours prior to arrival, which helped the AI system build a COVID-19 infection risk profile for all visitors. It collected information about travel history and other demographic data specific to the passenger. Eva then constructs the risk profile of arriving travellers by comparing them to the stored data of all previously tested passengers. The testing allocation is entirely based on an algorithm that balances two factors. Firstly, Eva determines the traveller to be at high risk for coronavirus infection based on past data. Second, if the available information is not enough to estimate the risk factor for an individual entering the country, all passengers tested based on the second criteria serve as the feedback channel, which helps Eva improve its predictions for all future travellers.
This allowed Greece authorities to test only a subset of travellers assessed by Eva as ‘high-risk’ groups and helped prevent asymptomatic COVID19-infected individuals from entering the country.
Effective screening methods are necessary to reopen borders
When South Africa detected a new coronavirus variant, named Omicron by the World Health Organisation, instead of being applauded for a vital discovery, many countries imposed travel restrictions on different African nations, especially South Africa. These kinds of measures will cause governments to suppress scientific findings relating to any new undiscovered strains due to the detrimental economic impact of travel bans.
Even though limiting non-essential international travel is necessary at critical times, developing more effective screening protocols will allow countries to reopen borders as soon as the situation is under control. To address this, AI-based screening methods such as Eva, the world’s first fully algorithmic reinforcement learning-based COVID-19 screening system, needs to be developed further and deployed throughout the globe. It will not only prevent coronavirus infection spread but also save valuable testing kits by limiting the number of travellers who have to undergo RT-PCR testing.
