Estimate of Covid-19 prevalence using imperfect data

Witold R Rudnicki; Radoslaw Piliszek

For most examined countries, with two exceptions, the mortality estimates obtained from all models is significantly larger than mortality for Germany. The exceptions are South Korea and Austria. South Korea by extensive testing and tracking of infections has managed to contain the epidemic and was able to quench the average number of new cases from nearly 800 in the beginning of March 2020, to less than 50 in the first ten days of April. Similarly, Austria performed extensive testing and apparently, as of April 10, managed to reduce the average number of cases from 800 at the end of March 2020, to around 300 at the time of the current study. Models 2 and 3 are in a good agreement with data for countries that are in the growth phase of the epidemic. In particular, it is remarkable that models with parameters optimised for German data fit extremely well to data collected for USA. Apparently, they adequately describe the convolution of distribution of probability from the test with rapid, nearly exponential, growth of number of new cases.