The coronavirus began to circulate a few weeks before the first cases were identified, and that we could put a name on the new disease. Scientists have looked for weak signals of this discreet emergence, investigating Twitter.
The SARS-CoV-2 coronavirus emerged in nature several weeks before being identified, most likely in early December 2019. The Covid-19 disease had therefore started to circulate before it was named, understood, or before the first cases are identified. In this in-between, people have been affected without knowing the novelty of the disease. Italian scientists from the IMT School for Advanced Studies Lucca had the idea to investigate the weak signals already present on social networks in December 2019 and January-February 2020.
In their research paper, published on January 25, 2021 in, they recount their investigation into the occurrences of terms related to symptoms of the coronavirus, within the messages published on Twitter.
The increase is significant
The authors were first interested in the word “pneumonia”. They created a database gathering hundreds of thousands of messages containing this term, and published in different languages (English, French, Italian, Spanish, German, Polish, Dutch). The database runs from December 2014 to March 2020, with the aim of detecting a development that would be truly significant, and not simply related to winter. All press articles and official messages, as well as retweets, were removed from the statistics taken into account by the study.
The results show that the occurrences of the word “pneumonia” increased significantly in most European countries over the winter of 2019-2020, compared to the same period in previous years. The increase is noticeable long before January 21, 2020 – when the disease was officially identified and named.
“” At the start of the pandemic, also note the scientists. For example, in France, the messages pointed mainly to Île-de-France; in Italy, in the region of Lombardy; in Spain, in Madrid. There is a link between the intensification of the occurrences of the word pneumonia and the most important epidemic foci.
The researchers came up with the same patterns of significant increase and geographic targeting for other symptoms caused by the SARS-CoV-2 coronavirus, such as “dry cough.”
Social networks, a precious tool?
This study published in does not claim that these weak signals could have detected the disease earlier. You cannot find what you are not looking for. Covid-19 disease was new. On the other hand, the results are quite relevant epidemiologically: they show how a pathogen can start to spread slowly in the population before being identified and before an exponential increase attracts attention.
“”, Say the researchers. They note, in conclusion of their study, that this type of investigation could henceforth make it possible to identify the emergence of new epidemic waves linked to the coronavirus.
The idea comes with its limits: “.”