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

Shin-ichiro Tanaka; Shinya Oku

Pearson's correlation coefficient r between ⟨ratios of accumulated positivity over population for 43 prefectures⟩ (%) and ⟨ratios of high-fever people over questionnaire responders for 43 prefectures⟩ (%) was calculated as r = 0.55. This means that the rate of PCR positivity and high-fever rate by LINE had an explicit correlation, showing the linearity with the slope of 6.35 and the intercept of 0.082 calculated by the least square method. Furthermore, the analysis with statsmodels library 8 proved that the slope values range 3.32-9.98 and the intercept values 0.074-0.090 with the significance level of 0.05. The lines with these values were displayed by dotted blue lines in Fig. 5 . As a note, if all the 47 prefectures were taken into the calculation, the correlation coefficient was as low as 0.28. With this value, while the use of data from all 47 prefectures seems inappropriate for the purpose of this paper, the authors could still confirm the relationship between these two ratios.