Forecasting Covid-19 Outbreak Progression in Italian Regions: A model based on neural network training from Chinese data

Abstract

Background. Epidemiological figures of Covid-19 epidemic in Italy are worse than those observed in China. Methods. We modeled the Covid-19 outbreak in Italian Regions vs. Lombardy to assess the epidemics progression and predict peaks of new daily infections and total cases by learning from the entire Chinese epidemiological dynamics. We trained an artificial neural network model, a modified autoencoder with Covid-19 Chinese data, to forecast epidemic curve of the different Italian regions, and use the susceptible-exposed-infected-removed (SEIR) compartment model to predict the spreading and peaks. We have estimated the basic reproduction number (R 0 ) -which represents the average number of people that can be infected by a person who has already acquired the infection -both by fitting the exponential growth rate of the infection across a 1-month period, and also by using a day by day assessment, based on single observations. Results. The expected peak of SEIR model for new daily cases was at the end of March at national level. The peak of overall positive cases is expected by April 11 th in Southern Italian Regions, a couple of days after that of Lombardy and Northern regions. According to our model, total confirmed cases in all Italy regions could reach 160,000 cases by April 30 th and stabilize at a plateau. Conclusions. Training neural networks on Chinese data and use the knowledge to forecast Italian spreading of Covid-19 has resulted in a good fit, measured with the mean average precision between official Italian data and the forecast.

Applications of artificial intelligence in battling against covid-19: A literature review

“…A Neural Network approach is presented in [391] which is a modified auto-encoder and is used to predict the epidemic curve of different regions in Italy. In [392] , an ANN is used on a publicly available dataset that contain information on infected, recovered and deceased patients.…”

Characterization of partially observed epidemics through Bayesian inference: application to COVID-19

“…, y (pf,m) } → p pf ( y pf | y). (18) Here, y (pf) denotes hypothetical data y and p pf ( y (pf) | y) denotes the push-forward probability density of the hypothetical data y (pf) conditioned on the observed data y. We start with samples from the posterior distribution p( | y).…”

“…The sampling workflow is similar to the one shown in Eq. (18). After the model evaluations y = n( ) are completed, we add random noise consistent with the likelihood model settings presented in Sect.…”

“…Our ignorance of its behavior in the human population is evident in the choice of modeling techniques used for the purpose. Time-series methods such as ARIMA [9,39] and logistic regression for cumulative time-series [28] have been used extensively, as have machine-learning methods using Long Short-Term Memory models [16,17] and autoencoders [18]. These approaches do not require any disease models and focus solely on fitting the data, daily or cumulative, of new cases as reported.…”

Forecasting Covid-19 Dynamics in Brazil: A Data Driven Approach

“…We discuss our results in comparison to those obtained with traditional methods and found that they work at least with the same precision in predicting the pandemics events. This paper follows a preliminary study that was carried out in Italy, based on China data with satisfactory results, but with space for improvement [ 9 , 15 ]. An approach based on Long Short Term Memory for Data Training (LSTM) has been initially tested by our team in Brazil, and demonstrated problems that are discussed in Section 2.2 .…”