COVID-SGIS: A smart tool for dynamic monitoring and temporal forecasting of Covid-19

This article has 1 evaluations Published on
Read the full article Related papers
This article on Sciety

Abstract

Objective

The new kind of coronavirus SARS-Cov2 spread to countries in all continents in the World. The coronavirus disease 2019 (Covid-19) causes fever, cough, sore throat, and in severe cases shortness of breath and death. To evaluate strategies, it is necessary to forecast the number of cases and deaths, in order to aid the stakeholders in the process of making decisions against the disease. We propose a system for real-time forecast of the cumulative cases of Covid-19 in Brazil.

Study Design

Monitoring of all Brazilian cities using oficial information from the National Notification System, from March to May 2020, concentrated on Brazil.io databases. Training and evaluation of ARIMA and other machine learning algorithms for temporal forecasting using correlation indexes (Pearson’s, Spearman’s, and Kendall’s) and RMSE(%). Validation from the relative errors of the following six days.

Methods

Our developed software, COVID-SGIS, captures information from the 26 states and the Distrito Federal at the Brazil.io database. From these data, ARIMA models are created for the accumulation of confirmed cases and death cases by Covid-19. Finally, six-day forecasts graphs are available for Brazil and for each of its federative units, separately, with a 95% CI. In addition to these predictions, the worst and best scenarios are also presented.

Results

ARIMA models were generated for Brazil and its 27 federative units. The states of Bahia, Maranhão, Piauí, Rio Grande do Norte and Amapá, Rondônia every day of the predictions were in the projection interval. The same happened to the states of Espírito Santo, Minas Gerais, Paraná and Santa Catarina. In Brazil, the percentage error between the predicted values and the actual values varied between 2.56% and 6.50%. For the days when the forecasts outside the prediction interval, the percentage errors in relation to the worst case scenario were below 5%. The states of Bahia, Maranhão, Piauí, Rio Grande do Norte, Amapá, and Rondônia every day of the predictions were in the projection interval. The same happened to the states of Espírito Santo, Minas Gerais, Paraná and Santa Catarina.

Conclusion

The proposed method for dynamic forecasting may be used to guide social policies and plan direct interventions in a robust, flexible and fast way. Since it is based on information from multiple databases, it can be adapted to the different realities, becoming an important tool to guide the course of politics and action against Covid-19 pandemic worldwide.

Related articles

Related articles are currently not available for this article.