It is a site well known to journalists, health professionals and the curious interested in real-time monitoring of the pandemic. Each month, 1.5 million visitors (including Emmanuel Macron) consult the graphs, figures and statistics published on the CovidTracker platform. Incidence rate, occupancy rate of intensive care beds, number of tests, vaccines, distribution of cases by department, age group… the site offers an enlightening summary of the pandemic situation, day by day.
“Our job is to make the data speak”, explains Elias Orphelin, one of the (young) figures experts involved in the case. To compare, to indicate the evolutions from one week to the next, to draw trends and projections. “We aggregate them, translate them, but harvesting them, we would be unable to do it ourselves. All the indicators come from the authorities, who do all the work, upstream. “
And the mass of data to be explored is great. “It’s like a gigantic Excel table, describes Elias Orphelin. We are juggling around ten source files. One of them has… 114,000 lines ”, calculate the one who is not yet in his thirties, a business school student during the day, datascientist (data expert) in the evening.
The idea of CovidTracker is not his. It germinated, at the very beginning of the first wave, with Guillaume Rozier, a 24-year-old computer engineer. At the time, the pandemic was approaching, but only raging in Italy. Public Health France does not yet publish the figures for the crisis, Jérôme Salomon, the Director General of Health, does not yet appear on television every night. To follow the evolution of the epidemic, Guillaume Rozier relies on data from the American Johns-Hopkins University and watches over the regional press. Its graphics catch the eye of Internet users. Elias Orphelin joined the team during confinement. “I wanted to give meaning to all this”, he remembers. Publishing public health data, making it intelligible, is “Help people see more clearly”, and “To make a contribution to the building”.
He remains impressed with what can be deduced from health data. “Knowing that your best friend has Covid doesn’t concern me. But if the data collected shows that his own neighbor is also affected, as well as his entire neighborhood, then we leave a particular case to achieve a more global vision. ” From there, it becomes possible to identify sources of contamination, to set up targeted strategies (local containments).
Passionate about big data, Guillaume Rozier, Elias Orphelin and their colleagues (around ten of them bring CovidTracker to life) do not let themselves be intoxicated by the figures. “The data we are handling is terrible. We never forget that behind a number of positive tests which climbs, behind an entry in more sheave, there are perhaps tragedies which are brewing. “