The approach of constructing correlation networks can be applied to countries, each of which is characterized by its own process of spreading the pandemic. Previously, it was shown that non-directional correlation networks of parameters allow defining clusters of objects. Non-directionality, on the one hand, reduces the quality of clustering, and on the other hand, it does not allow us to get closer to the problem of finding causal relationships. A model of correlation networks is therefore proposed by the authors, which takes into account the absolute values of the compared measurement series and the mutual offsets of these series. As a result of the implementation of the model, in part, directional correlation networks are formed, determined by the dynamics of COVID-19 distribution in various countries. The paper shows that node sizes, link weights, and the clustering of such networks leads to easily interpreted results. The proposed methodology can be used both to study the spread of the pandemic in various countries and to study other social, political and economic processes. Keywords: COVID-19, pandemic, correlation networks, directed networks, datasets, new incidents of infection. Distributed in Computational Biology eJournal Vol 4, Issue 44, September 16, 2020. Distributed in Human Health & Disease eJournal Vol 4, Issue 129, September 14, 2020. Distributed in Coronavirus & Infectious Disease Research eJournal Vol 1, Issue 129, September 15, 2020. Distributed in Public Health eJournal Vol 2, Issue 138, September 01, 2020. Distributed in Infectious Diseases eJournal Vol 2, Issue 108, August 31, 2020.