I Spent My Summer Using AI To Help Save Greece from COVID-19

It was three months ago when my friend Kimonas asked me if I can help him with a huge secret project that he had on his mind. It was kind of a cheap shot as “huge secret project” are my trigger words.

He asked me if I can join him on a Zoom at 6am. I told him that I am not going to wake up that early even if the President of Greece was on that Zoom call.

It turned out that the Prime Minister and his team of scientists were on the call and I was there, 7am in Los Angeles, half awake, wearing my “A.I pays my bills” t-shirt. The “huge secret project” was not clickbait.

We are going to build something that has never been built before.

If you don’t know Greece, I assume it is because you hate good food, sunshine, breathtaking beaches, history, and you haven’t watched “My Big Greek Fat Wedding”. And this comes from someone that is 100% impartial on this (being Greek doesn’t mean I cannot be impartial, right?).

Greece has a population of 11 million and during non-COVID19 summers, the country hosts more than 33 million tourists. This is more than 3 times the population of the entire country. They all come to Greece and spend most of their time on the (tiny) Greek Islands. Oh, and most of them come during the three (summer) months.

Tourism also supports nearly a quarter of the country’s jobs and represents 20% of the nation’s GDP.

In the context of COVID19, this is a double-edged sword for the country. On one hand, keeping its borders closed to tourists equals financial suicide for Greece — a country that had just started to recover from a previous financial hardship. On the other hand, blindly accepting 100K to 300K tourists PER DAY, translates to “Death by COVID19”, by putting the locals at risk of infection. To the best of our knowledge, there is no country that has the capacity to conduct these many tests or this kind of infrastructure or solution to pull this off.

Until now.

Meet EVA

EVA is a country-wide AI screening and testing program , thought to be the first of its kind in the world, which has been online since July 1st 2020.

Here is how it works.

Let’s say you are one of the lucky ones that can work remotely. What better place to work from than the seaside, munching on tasty Mediterranean food. Again, 100% impartial on this.

Before your arrival in Greece, you must fill out an online “passenger locator form” (we call it PLF) providing basic information including which country they are coming from, gender, age, and other details.

After submitting your information to the GDPR-compliant system (Europe is extremely careful regarding how personal data is stored and handled), a subset of anonymized data arrives to EVA. EVA takes into account how many other people will arrive at the same entry point, availability of tests, and your personal risk profile and responds with a “TEST” or “NO TEST”. All in realtime.

This “test/no-test” information is sent back to you via email and encoded as a QR CODE that will be scanned upon your arrival in Greece.

The immigration officer will know at the time of scanning whether you should be sent for testing or you are free to continue your vacations.

If you are sent for testing, your (anonymized) lab results will be fed back to EVA to improve the model and its accuracy — closing the self-learning loop.

By digitizing the details of a visitor’s trip, Greece can make informed decisions on who should be tested, how tests should be allocated, or whether any borders need to be closed due to increasing risk.

How good EVA is?

My personal opinion (using only public data) is that EVA is working.

Currently Greece has less than 20 new external infections from 100k visitors per day. The total infections are on average 150 to 200 mostly to internal spread of the virus which is extremely low.

I don’t know any other country that managed to stay open for as long as Greece while having thousands of tourists per day and open bars and restaurants. Of course if Greece wants to keep infections low, will have to impose stricter controls to manage “internal infections”.

EVA might seem like a “Deus Ex Machina” but as Greeks say

You cannot expect everything from the Gods — you also have to put some work.

TeamEVA

EVA is the brain child of three very smart people. And me.

The team is:

• Hamsa Bastani, assistant professor of operations, information and decisions from the Wharton School at the University of Pennsylvania
• Kimon Drakopoulos, assistant professors of data science and operations at the Marshall School
• Vishal Gupta, assistant professors of data science and operations at the Marshall School
• And me

The whole team is committed in open-sourcing the important parts of EVA and of course a full publication will be coming up in December 2020.

You can find more information if you follow these links:

• USC Marshall, Wharton Researchers Team with Entrepreneurs and Policy Makers to Help Reopen Greek Economy
• LDI/Wharton-USC Collaborative Plays Key Role in Greece’s COVID-19 Response
• A Data-Driven Reopening

Also published here.