Um estudo aplicado no estado de São Paulo utilizando Redes Bayesianas na predição do controle no avanço de COVID-19

Abstract

Since december 2019, our planet has been suffering from an unexpected pandemic of coronavirus 2019 (COVID-19), due to severe respiratory sindrome caused by coronavius 2 (SARS-CoV-2). Since it trasmits by droplets from cough or sneeze and with physical contact with infected people resulted in aproximately 1 million of death all around the world. Patients' clinical and epidemiological features of COVID-19 are being informed, and a diversity of systems are being developed to diagnose this virus to contain its advance. In this study, we propose and simulate a predictive model based on a Bayesian Network to model the classification used bay Sao Paulo state to contain the virus, which is based on the health system and epidemic advance. Furthermore, a Bayesian Network based on clinical and epidemiological data off COVID-19 was developed, thus to verify probabilities and correlations of death index given analyzed features. The models weredeveloped using existing data available by SEADE, Health Ministery and Plano SP, thus, being possible to infer befitting decisions made by Plano SP and it is added a flexibilization node to be used by a managing unit. Moreover, the second Bayesian Network was possible to observe a probabilistic relationship between risk factors and number of deaths in Sao Paulo state.