DOI: https://doi.org/10.20535/kpi-sn.2020.2.205127

THE FORECAST OF CONFIRMED COVID-19 INFECTION CASES IN UKRAINE

Olena M. Pavliuk, Natalia K. Lysa, Olga Yu. Fedevych, Anastasiia-Olha A. Strontsitska

Abstract


Background. COVID-19 pandemic is one of the world’s newest and most actual problem. The only workable way to stop the spreading of the decease is to introduce quarantine. The stronger quarantine restrictions the stronger they affect the economic and social situation.

Objective. Based on the analysis of the literature sources and the own research, we forecast the number of patients with COVID-19 in Ukraine and develop recommendations on the feasibility of introducing the next stages of quarantine limitations.

Methods. The following methods were used: the correlation identification between the number of patients with coronavirus in Ukraine and other countries; the development of the trends for long-term and medium-term forecasts of the number of patients with COVID-19 in Ukraine; the use of the non-iterative artificial neural networks (ANN) based on the radial basis functions (RBF) with additional lateral connections between neurons of the latent layer to make a short-term forecast of the number of patients with the decease.

Results. The correlation between the number of coronavirus patients in Ukraine, Italy and Spain was revealed. The trend was constructed with the help of polynomials from the 2nd to the 6th degree for long-term forecast of the number of patients with COVID-19 in Ukraine and the long-term forecast was made. The similarities of Ukraine's trajectories with Poland and Sweden were used for the medium-term forecast. It is established that Ukraine is increasingly deviating from the trajectory of Poland and approaching the trajectory of Sweden with a delay of 2-3 days. ANN RBF with additional lateral connections between neurons of the latent layer were used for short-term forecast of the number of patients with COVID-19 in Ukraine. The RMS error of the ANN training for the 14-day forecast is 0.55%, and the maximum is 1.43%.

Conclusions. Based on the forecasts, the effect of the first stage of quarantine easing is shown, and the second stage of quarantine restrictions showed the effect on the disease.


Keywords


Number of patients; COVID-19; Pandemic; SARS-CoV-2 virus; Quarantine; Short-term forecast; Middle-term forecast; Long-term forecast; Non-iterative artificial neural networks; Radial basis functions

References


Scientists from Oxford Eestimate that There are Almost 13 Thousand Patients with the Coronavirus in Ukraine [Online]. Available: https://glavcom.ua/news/vcheni-z-oksforda-vvazhayut-shcho-v-ukrajini-mayzhe-13-tis-hvorih-na-koronavirus-670592.html

Project FORSITE COVID-19 of the World Data Center for Geoinformatics and Sustainable Development, Igor Sikorsky KPI [Online]. Available: http://wdc.org.ua/uk/node/190016

End of the Covid-19 Pandemic in Ukraine: Scientists Made an Optimistic Forecast [Online].

Available: https://prm.ua/zavershennya-pandemiyi-covid-19-v-ukrayini-vcheni-zrobili-optimistichniy-prognoz/

Forecast of the Development of the COVID-19 Epidemic in Ukraine in the Period from April 28 to May 5, 2020 and Analysis of Possible Scenarios for the Development of the Epidemic in the Period from May 6 to June 5, 2020. “Forecast of WG-3” [Online]. Available: http://files.nas.gov.ua/PublicMessages/Documents/0/2020/04/200428221156130-7671.pdf

R. Tkachenko et al., “Committee of the combined RBF-SGTM neural-like structures for prediction tasks”, Lecture Notes Comp. Sci., vol. 11673, pp. 267–277, 2019. doi: 10.1007/978-3-030-27192-3_21

I. Izonin et al., “Committee of SGTM neural-like structures with RBF kernel for insurance cost prediction task”, in Proc. IEEE 2nd Ukraine Conference on Electrical and Computer Engineering, 2019, pp. 1037–1040. doi: 10.1109/ukrcon.2019.8879905

R. Tkachenko et al., “Committee of the SGTM neural-like structures with extended inputs for predictive analytics in insurance”, Commun. Comp. Inform. Sci., vol. 1054, pp. 121–132, 2019. doi: 10.1007/978-3-030-27355-2_9


GOST Style Citations




Refbacks

  • There are currently no refbacks.




Copyright (c) 2020 The Author(s)

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.