Statistical and network-based analysis of Italian COVID-19 data: communities detection and temporal evolution

Marianna Milano; Mario Cannataro

other regions. Next, in the network resulting by considering the all weeks, Lombardia, Veneto, Emilia and Marche formed a subgraph disconnected by a different dense subgraph composed by the rest of the regions. However, Veneto represents a single community in Figure 12 (d). This means that the behaviour of Veneto presents a low similarity respect to Lombardia, Emilia and Marche despite forming a module. It is possible to notice that: Lombardia, Veneto, Emilia form a community in both Total Currently Positive network and New Currently Positive network, Piemonte represent in sigle community in Total Currently Positive network and while Piemonte is associated with Marche and Toscana in New Currently Positive network. Figure 8 represents the Discharged/ Healed network over the five weeks. In the first week, the network structure present all nodes connected with exception of Veneto. In the second week, Discharged/ Healed network is formed by three subgraphs; in the third and fourth weeks the network results very dense; in the fifth week is caracterized by different disconnected components, and finally, at the end of 35 days the network is composed by a subgraph composed by Lombardia and Veneto and another subgraph highly connected. This means that Lombardia and Veneto have a similar behaviour that it is different by the rest of the Italian regions. Also Lombardia and Veneto represents one of the five communities extracted by Discharged/ Healed network. The extracted communities are reported in Figure 12 (g). Figure 9 shows the evolution of Deceased network. The evolution of this network is different by other Italian COVID-19 networks. In fact, in first week all nodes are disconnected, so all Italian regions present 7 All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.