Researchers from the Institute for High-Performance Computing and Networking (ICAR) of the CNR, in collaboration with colleagues from the DIEM of the University of Salerno, have designed and implemented a forecasting tool for the epidemiology of Covid-19.
This tool, based on the well-known SEIR epidemiological model, has been improved by providing it with two important additional features.
Firstly, the model has been made more suitable for the real situation, by adding to it a characteristic of social distancing that can vary over time, so as to be able to take into account the various decrees promulgated by the Government as well as their effects in terms of restrictions on mobility.
Secondly, one of the problems for an efficient use of the SEIR model is represented by the choice of the values of its parameters. This choice has been done by combining the model with an optimization algorithm from Artificial Intelligence, more precisely from Machine Learning.
By using this forecasting tool, at the end of March the epidemiological trend for Covid-19 was predicted for Italy, and for its Campania and Lombardy regions, starting from the data available up to that day. This was done in terms of forecasting the daily number of new positive cases. The results were presented on April 1 in a publication made available online in the freely accessible ArXiv scientific archive.
In particular, for the Campania region, it was forecast on April 1st that the number of new positive cases “approaches zero around June 3”.
The most recent official data for Campania is: June 3: 1 new positive case; June 4: 0 new positive cases.
For more information, this paper is freely accessible online in ArXiv at the address: https://arxiv.org/pdf/2004.00553.pdf