Sobre o Problema de Previsão da Covid-19 Utilizando Modelos Morfológico-Linear Profundos

Abstract

The coronavirus disease 2019 (COVID-19) has been declared by the World Health Organization (WHO) as an unprecedented pandemic in the present days, straining healthcare systems due to the high demand for admissions to intensive care units. In this context, estimating the dynamics of the COVID19 pandemic is essential to deal with health system drawbacks. Therefore, in this work we developed an empirical study on time series related to the COVID-19 pandemic and, based on this study, we present a deep morphological-linear model, trained by a gradient-based learning process, able to predict this particular kind of time series. Trying to assess the predictive performance of the proposed model, we use daily COVID-19 time series in Brazil and United States of America. The achieved results show that the proposed model outperforms classical and recent machine learning models to estimate the dynamics of the COVID-19 pandemic.