At the end of 2019, a new disease called COVID-19 hit the entire world. Firstly identified in Wuhan (China), it spread around the globe, infecting and killing people belonging to the most diverse spheres. However, it did not affect all regions in the same way. Therefore, the goal of this research is to understand why Lombardy has been so drastically hit during the first two waves of the outbreak, investigating the external contributing factors that may have affected the density of COVID-19 cases, and the effectiveness of the anti-contagion policies employed in Europe.
To achieve this objective, an aggregated open-source dataset was created, to include a heterogeneous set of twenty-two variables, related to six macro areas of interest: economy, healthcare, population, primary sector, mobility and education. This dataset, together with machine learning techniques, revealed that the same external contributing factors, that were important for the prediction of the risk-category of COVID-19 density for a specific region, were also the ones displaying Lombardy as an outlier when compared to other European regions. The main predictive contributing factors were life expectancy (positively correlated), the amount of working hours (positive correlation).
On the other hand, the analysis of the policies and their effectiveness was based on the data available in the CoronaNet dataset, which collects the worldwide application of non-pharmacological regulations.
The application of statistical descriptive methods computed the effectiveness of each policy type, by comparing the contagion curve trend between when the policy is enacted and 7-20 days after, when its effects should become evident. These analyses showed that the restriction of businesses and interregional transit and the enactment of lockdowns have been the most impactful policies. It is important to underline that these results must be considered as a general indication, since the implementation of similar policies can be enacted with varying degrees of strictness, which are not captured by the considered dataset.
Finally, since this pandemic revealed the need for accurate data representation, these analyses are published on a public website, named CovidatLombardy, available to anyone who is interested in the topic. The website will not only provide a detailed description of the analysis and its results, but it will also allow the audience to freely navigate the data through interactive maps. The website will have both an author-driven and a reader-driven approach, creating a meeting point between scientific and information visualisations.
Principal Academic Tutor:
Stefano Ceri (Politecnico di Milano)
Elena Baralis (Politecnico di Torino)
Fabio Pammolli (Politecnico di Milano)
Prof. Ilaria Capua – University of Florida