You can use any machine learning algorithm for forecasting the corona trends globally or country-specific. You can find few things here,

https://www.technologyreview.com/s/615360/cdc-cmu-forecasts-coronavirus-spread/

https://www.teradata.com/Blogs/Advanced-Analytics-for-Coronavirus-Trends-Patterns-Predictions

https://towardsdatascience.com/using-kalman-filter-to-predict-corona-virus-spread-72d91b74cc8

https://www.medrxiv.org/content/10.1101/2020.01.23.20018549v1.full.pdf

https://www.medrxiv.org/content/10.1101/2020.03.21.20040154v1.full.pdf

Thank you very much @christian

This is challenging, because the number of confirmed cases are already lower limits, because you can expect that there is much more cases than recoded. This (should) influences also your forecasts. For the deaths its also complicated, because the data quality varies from country to country. The basic question is, is somebody died from or with Corona. Thus the quality of the observed records is weak, and conventional forecasting methods may not consider this.

An option for data scientists who can contribute to the medical community? https://towardsdatascience.com/covid-19-how-data-scientists-can-contribute-to-the-medical-community-ad03b19d34f0

A significant part of currently used prognostic models for analyzing the development of a pandemic (or other multifactorial, consisting of large data sets of issues) is based on Big Data Analytics analysis regarding quantitative historical data on the number of people infected with SARS-CoV-2 Coronavirus (causing Covid-19 disease), including people who have been diagnosed with this virus in the body on the basis of tests carried out, people suffering from Covid-19 disease, people who have died from this disease (and to a certain extent from the presence of other diseases and co-morbidities in the body) and people who have high the probability level is or were infected with SARS-CoV-2 Coronavirus, but they did not know it (currently this part of the population may already be relatively the largest in the pandemic countries). When constructing prognostic models, the key assumptions of impact factors and thus creating projections for the development and / or expiry of a pandemic are formulated on the basis of current knowledge in the field of virology and immunology, sanitary and public safety, etc., including current information on therapeutic therapies used (e.g. based on blood plasma containing naturally produced antibodies), vaccine development prospects, anti-pandemic and sanitary safety instruments used (e.g. home quarantine, restrictions on movement in public places, wearing protective masks, frequent hand decontamination, etc.) and the results of previous tests and analyzes of the effectiveness of these key determinants mentioned above. In view of the above, it is important to constantly improve the forecasting models built, which should have a dynamic formula, constantly updated.

Regards,

Dariusz Prokopowicz