In January 2020 the coronavirus was already circulating … on social networks. According to an Italian study published on January 25, 2021 in the journal Nature Scientific Reports, Twitter posts reflect an increase in concerns about pneumonia in Europe between late 2019 and early 2020.

Behind “pneumonia”, the start of covid

For this study, economics and statistics researchers at the IMT School for Advanced Studies in Lucca, Italy, created a database of all tweets containing the keyword “pneumonia“from December 2014 to March 1, 2020 in the seven most widely spoken languages ​​in Europe: English, German, French, Italian, Spanish, Polish and Dutch.

Why this keyword? Because it is a possible evolution of a coronavirus infection. And since the 2020 flu season has been relatively mild, it was likely the source of few lung complications. Who said “pneumonia“in January 2020 therefore probably said”covid” without knowing.

Lombardy, Madrid, Ile-de-France

Result? As of January 2020, the number of tweets including the keyword “pneumonia“is increasing in most European countries. In Italy, for example, where the first containment measures started on February 22, 2020, there are more mentions of pneumonia in the first weeks of 2020 than in the same period, in 2019.

Better still, by geolocating 13,000 tweets, the researchers noticed that the first warning signals came exactly from the regions where the first outbreaks were then detected: Lombardy in Italy, Madrid for Spain and Île-de-France. for France.

Thus, for the example of Lombardy, the tweets testified to the presence of cases from early January, several weeks before the announcement of the first local infection of covid and the first containment measures initiated on February 20 in this region. .

A monitoring tool?

Our study is further evidence that social networks can be a useful tool for epidemiological surveillance “ explains Massimo Riccaboni, professor of economics and co-author of the study in a IMT press release. “They can help catch the first signs of a new disease, before it silently grows, and also track its spread.“he continues.

This is all the more true for covid-19, for which early warning signals are still poorly identified. In the months and years to come, monitoring of social networks could therefore, according to the authors of the study, constitute a precious help for the health authorities to reduce the risks of resurgence. Either by locally adopting strict social distancing measures in the event of an alert, or on the contrary by mitigating them in other quieter regions.