DEEP LEARNING MODEL FOR THE PREDICTION OF COVID-19

CHEUTEU TAZOPAP JOSEPH ROMARIC

Keywords: Covid-19, Deep Learning, LSTM, Mathematical MODELING

TL;DR: Prediction of covid-19 using Mathematical MODELING and Deep Learning model

Abstract:

Covid-19 is an infectious disease caused by a strain of Coronavirus called SARCOV-2. This disease
started in China on November 16, 2019 in Wuhan in Hubei province and was declared by the World
Health Organization as a state of emergency on March 11, 2020. This epidemic started in Cameroon on
March 14, 2020 in the far north region and up to our the whole world has experienced over 650 million
confirmed cases with over 7 million deaths. This work concerns mathematical modeling of COVID-19
transmission in Cameroon using a deep learning. We establish first a mathematical model that
describes the transmission of COVID-19 within the Cameroonian population using time-depending
parameters and known constant parameters. We present a theoretical analysis of the model, more
precisely, we prove the existence, uniqueness, the positivity, the boundedness of the solutions. We also
determine the points of equilibrium. Secondly, we establish two deep learning models, namely an LSTM
(Long Short Term Memory) and a GRU (Gated Recurrent Unit) and then we make the identification of
the time-dependent parameters thanks to the daily data that we have had and during the training, the
test and the predictions we obtain the best scores with the coefficient of determination which is
between 0.96 and 0.98. Then we will extend our study by studying several other digital intelligence
models such as machine learning models and we will also be able to adjust our mathematical model
of transmission of COVID-19 in Cameroon and apply our model in several other countries.

Submission Category: Machine learning algorithms

Submission Number: 41