A Network Modeling and Analysis of COVID-19 Hospital Patient Data

Abstract

There are currently large amounts of public databases on COVID-19 patients. Among them the FAPESP COVID-19 Data Sharing/BR has gathered laboratory tests and hospitalization data from large health centers in the São Paulo metropolitan area. This paper uses part of such a repository to assemble a set of networks of positive COVID-19 patients according to the similarity of their age and laboratory tests results. Next, popular complex network metrics such as clustering coefficient and average path length are extracted from such networks and compared to the expected values observed for classical networks. Similarities of the clustering coefficient values with those of Watts-Strogatz networks were observed, although there are no sustainable characteristics of Small World networks. There are also similarities to scale-free networks, such as high-degree variance and the presence of hubs of nodes. In addition, a partition of the networks based on the modularity measure using the Fast Greedy algorithm is obtained and analyzed. An structure with four clusters and modularity values greater than zero was found, indicating that the network has some community structure.

Filipe A. N. Verri

Researcher

My research interests include data science, machine learning, complex networks, and complex systems.