Estimating Parameters of a SEIRD Model Applied to SARS-CoV-2 Infections in Germany based on the Particle Swarm Optimization Method

Authors

Evandro A. Nakajima, Universidade Tecnolo ́gica Federal do Parana ́ (UTFPR), Campus Santa Helena, PR, BrazilFollow
Antonio A. Ignacio, Universidade Tecnolo ́gica Federal do Parana ́ (UTFPR), Campus Santa Helena, PR, BrazilFollow
Denise Lange, Universidade Tecnolo ́gica Federal do Parana ́ (UTFPR), Campus Santa Helena, PR, BrazilFollow
Erika Izumi, Universidade Tecnolo ́gica Federal do Parana ́ (UTFPR), Campus Santa Helena, PR, BrazilFollow

Author Country (or Countries)

Brazil

Abstract

The present paper describes a particle swarm optimization (PSO) method used to estimate parameters of a Susceptible- Exposed-Infected-Recovered-Dead (SEIRD) model applied to predict SARS-CoV-2 transmission in Germany, based on data from February 15th to April 25th, 2020, considering that the lockdown in the country started on March 23rd. The model estimated patients’ mortality (4.92%) and recovery rates (95.08%), virus incubation (8.54 days), infection periods (18.65 days), as well as the basic virus reproduction number before (R0 = 11.60) and after (R0 = 0.39) lockdown. The predicted values were accurate until the 70th day. The performances achieved by the model were 0.98 for infected, 0.97 for the recovered and 0.97 for the dead, asserting the model’s great performance (> 0.75). The model also suggests that on February 15th, 2020, there were 67 infected individuals in the incubation period. We believe that this model can help other studies to better understand and accurately predict epidemic curves, mainly in countries where the new coronavirus has recently started to spread. It also may guide public health policies that aim to control the disease.

Digital Object Identifier (DOI)

http://dx.doi.org/10.18576/amis/150403

Recommended Citation

A. Nakajima, Evandro; A. Ignacio, Antonio; Lange, Denise; and Izumi, Erika (2021) "Estimating Parameters of a SEIRD Model Applied to SARS-CoV-2 Infections in Germany based on the Particle Swarm Optimization Method," Applied Mathematics & Information Sciences: Vol. 15: Iss. 4, Article 3.
DOI: http://dx.doi.org/10.18576/amis/150403
Available at: https://digitalcommons.aaru.edu.jo/amis/vol15/iss4/3