Had been seen since December the arrival of the new ‘sister’ variant of Omicron, the sub-variant BA.2, thanks to mathematical tools able to go so deep into the data analysis, to be able to see anomalies otherwise not detectable: “the presence in Italy of the new variant that was spreading from the northern regions had already been identified a few weeks ago, much earlier of the identification of the genetic sequenceStatistician Livio Fenga, who recently moved from Italy to Great Britain, where he is senior lecturer of the Center for Analysis, Simulation and Models of the British University of Exeter, told ANSA. “Analysis tools like these – he added – are a lot important to succeed in anticipate the arrival of the new variants“.
To give an idea of how they work, let’s think of the electricity consumption recorded in an office where routine work is carried out (photocopiers, faxes, telephones, air conditioning). Now, “let’s suppose that once a week, from 15.00 to 16.00, all the printers have to work at the same time for work needs: an anomalous situation like this generates shocks that are not easy to detect based on the observation of the historical series of electricity consumption (technically speaking of time – amplitude domain). Now, we can compare the onset of a new variant as something that is added (printers 3 to 4) to the chore (the dynamic that describes the pre-variant situation).
These effects could of course be the result of phenomena other than those of a variant, for which a great deal of prudence is always required in the interpretation of the phenomena associated with Covid.
Detect such anomalies it is far from simple: very often they are very weak signals and to be able to detect its presence, the so-called ‘ghosts‘, a kind of mathematical filters that allow you to highlight in the data peaks occurring only at certain frequencies.
“TO December, for example, the analysis of spectra applied to Italian data on the pandemic highlighted gods anomalous peaks, interpreted as a first warning signal of the presence of a new variant “.
This does not mean that the variant was present then: predictions of this type, called ‘nowcast‘, “Allow you to look now at all the data that could be seen in the future. Using such a model in December it was possible to detect the possibility of a new variant coming to light, ”Fenga observes.
But analyzing the spectra is not enough: a second tool was also used for the prediction, which refers to the theory of Wavelet, literally ‘undines’, small oscillations whose behavior allows us to analyze complex phenomena, such as meteorological ones. “This tool allows you to analyze anomalies in detail, like a sort of zoom capable of magnifying the variations of the signal at a certain frequency. “They are tools well known to statisticians – concluded Fenga – and could prove to be very useful in the future, to anticipate the arrival of new variants”