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

Shin-ichiro Tanaka; Shinya Oku

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.15.20066100 doi: medRxiv preprint With this consideration above, the authors conducted an additional analysis using the data investigated and made open by J.A.G. Japan 5 displayed in Fig. 2 . This dataset has the data of initial symptom and those of confirmed positivity (as of April 12). As the date of initial symptom was not identified for every patient, this data showed only the identified data, meanwhile, satisfactorily statistical distribution was exhibited. The authors superimposed the Gaussian distribution onto this data (where x is the difference between dates of initial symptom and confirmed positivity). Although this distribution model was not completely concordant with the data and there might exist some other better distribution which fits more, it was shown that there are 4 to 5 days of difference between the dates of initial symptom and confirmed positivity. It could be reasonable to use the confirmed positivity data on April 5 to be compared with LINE questionnaire data on March 31 to April 1. Fig. 3 displayed the accumulated number of confirmed positivity on April 5 per prefecture, divided by the population obtained from the National Survey 2015. It shows the fact that the rates of COVID-19 positivity varies largely with each prefecture. As the next step, the authors investigated the correlation between the distribution of positivity rate per prefecture and that of highfever rate. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.