Estimation of true number of COVID-19 infected people in Japan using LINE questionnaire

Shin-ichiro Tanaka; Shinya Oku

Here, the half-life is set as 9 day, which brought slightly improving regression coefficient, from 0.55 to 0.59 showing that the hypothesis applied here are reasonable to a large extent. As the change of patient number from accumulated to "non-accumulated" downwards, the estimation for the true number of the infected, adding consideration of half-life, was 23,783, producing no outstanding difference from the number of the precedent method. Despite this observation, with the increase of correlation coefficient and decrease of statistical error, the new range was 13,522-34,044 with the same significance level of 0.05, resulting in a slight improvement of precision. To confirm the appropriateness of half-life being 9 day, the authors added some other evaluations using 3 days and 14 days for half-lives; the results were 0.57 and 0.58 respectively and became slightly worse than employing 9 day half-life.