Back

Monitoring and forecasting the number of reported and unreported cases of the COVID-19 epidemic in Brazil using Particle Filter

DUTRA, J. C. S.; da Silva, W. B.; da Costa, J. M. J.

2020-05-29 epidemiology
10.1101/2020.05.27.20115212 medRxiv
Show abstract

In this paper, we combine algorithm of Liu & West for the Particle Filter (PF) with SIRU-type epidemic model to monitor and forecast cases of Covid-19 in Brazil from February up to September. We filter the number of cumulative reported cases and estimate model parameters and more importantly unreported infectious cases (asymptomatic and symptomatic infectious individuals). The parameters under study are related to the attenuation factor of the transmission rate and the fraction of asymptomatic infectious becoming reported as symptomatic infectious. Initially, the problem is analysed through Particle Swarm Optimization (PSO) based simulations to provide initial guesses, which are then refined by means of PF simulations. Subsequently, two additional steps are performed to verify the capability of the adjusted model to predict and forecast new cases. According to the results, the pandemic peak is expected to take place in mid-June 2020 with about 25,000 news cases per day. As medical and hospital resources are limited, this result shows that public health interventions are essential and should not be relaxed prematurely, so that the coronavirus pandemic is controlled and conditions are available for the treatment of the most severe cases.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.