Fitting two simple probabilistic models for covid–19 with data from Ecuador

Abstract:

Background: In this study, we use public data of COVID–19 diagnosed cases from Guayas province (Ecuador) and use them to evaluate the forecasting ability of some probabilistic models commonly used in the epidemiological and mathematical literature. We highlight the precision of the models in pbkp_redicting COVID–19 cases, and the evolution of the epidemic curve until 2021. Those models are helpful to estimate a possible stabilization date for the epidemic. Methods: We use Gompertz and Hill functions to provide robust pbkp_redictions of the COVID–19 diagnosed cases and the stabilization date of the epidemic curve. We simulate the model errors for pbkp_redictions by including a probabilistic error term from a Gamma distribution. Non–linear least square method is performed trough the Gauss–Newton algorithm with 5000 iterations to estimate each parameter in the models. The Coefficient of Determination, the Akaike Information Criteria and the Root Mean Squared Error are used as the metrics to evaluate the model perfor-mances. Conclusions: The accuracy of our mathematical models is high compared to diagnosed cases until December 9, 2020. Medium terms pbkp_redictions indicate an increase of cumulative COVID–19 cases until 2021. Since to date, the diagnosed cases will be increasing until the stabilization is achieved. According to the Hill model the stabilization will be achieved on February, 2021, while for the Gompertz model the stabilization will be on December, 2020.

Año de publicación:

2020

Keywords:

• SARS–CoV–2
• Hill model
• Gompertz model
• Pbkp_redictions. 1

Fuente:

scopus