Open Access

ARTICLE

Data Driven Modelling of Coronavirus Spread in Spain

by G. N. Baltas, F. A. Prieto, M. Frantzi, C. R. Garcia-Alonso, P. Rodriguez

1 Loyola Institute of Science and Technology, Universidad Loyola Andalucía, Seville, 41704, Spain.
2 Department of Biomarker Research, Mosaiques Diagnostics GmbH, Hannover, 30659, Germany.
3 Renewable Electrical Energy Systems, Technical University of Catalonia, Terrassa, 08222, Spain.

* Corresponding Author: Gregory N. Baltas. Email: .

(This article belongs to the Special Issue: Artificial Intelligence and Information Technologies for COVID-19)

Computers, Materials & Continua 2020, 64(3), 1343-1357. https://doi.org/10.32604/cmc.2020.011243

Received 28 April 2020; Accepted 21 May 2020; Issue published 30 June 2020

Abstract

During the late months of last year, a novel coronavirus was detected in Hubei, China. The virus, since then, has spread all across the globe forcing Word Health Organization (WHO) to declare COVID-19 outbreak a pandemic. In Spain, the virus started infecting the country slowly until rapid growth of infected people occurred in Madrid, Barcelona and other major cities. The government in an attempt to stop the rapssid spread of the virus and ensure that health system will not reach its capacity, implement strict measures by putting the entire country in quarantine. The duration of these measures, depends on the evolution of the virus in Spain. In this study, a Deep Neural Network approach using Monte Carlo is proposed for generating a database to train networks for estimating the optimal parameters of a SIR epidemiology model. The number of total infected people as of April 7 in Spain is considered as input to the Deep Neural Network. The adaptability of the model was evaluated using the latest data upon completion of this paper, i.e., April 14. The date range for the peak of infected people (i.e., active cases) based on the new information is estimated to be within 74 to 109 days after the first recorded case of COVID-19 in Spain. In addition, a curve fitting measure based on the squared Euclidean distance indicates that according to the current data the peak might occur before the 86th day. Collectively, Deep Neural Networks have proven accurate and useful tools in handling big epidemiological data and for peak prediction estimates.

---

Keywords

---