Contact-Tracing based on Time-Varying Graph Analysis

The widespread diffusion of the SARS-CoV2 virus in the last months has forced many organizations in different socio-economic fields to study new technologies to counteract its presence. One of these technologies is contact tracing based on the collection of device interactions by using specific apps and device capabilities. By tracing and storing device interactions, when a person is revealed as infected after a specific test, other people who entered in contact with the infected one are notified for an early screening aimed at stopping the infection spreading. Several solutions have been developed at regional, national or continental level, according to different architectures (centralized, decentralized and hybrid) depending on the degree of desired privacy. However, none of them exploits interaction tracing to build graphs for early detection of critical identities. In this paper, we propose an architecture, a framework and an algorithm to identify, in quasi-real-time, critical spots in time-varying graphs inferred from device interactions captured by scanning Bluetooth advertisements. We show, by examples, how the approach could provide useful information for early detecting critical people in order to prioritize mass screening for improving the effectiveness and breaking infection chains. Finally, a prototype implementation is presented.