Comment on “Predicting the mortality due to Covid-19 by the next month for Italy, Iran and South Korea; a simulation study”
Gastroenterology and Hepatology from Bed to Bench,
20 September 2020
We have read with great interest the manuscript by which uses statistical models to predict the mortality due to Covid-19 for Italy, Iran and South Korea for the next month from 3/15/2020. It was claimed that the distribution of mortality was Poisson so the corresponding methods were used for performing simulations . Now the actual values of mortality have released by WHO for the prediction period and the simulations showed an acceptable prediction but we would like to raise some comments.
Figure 1 shows the Rate ratios (actual rate/prediction rate) in Italy, South Korea and Iran. Rate ratios shows an overestimation of mortality rate in Italy (0.78, 0.65, 0.77) and South Korea (0.83, 0.77, 0.83) but an underestimation in Iran (1, 126.96.36.199). The best predictions as expected were duo to next 10-Day of forecast. Although the same pattern of errors in predictions were occurred for Italy and South Korea, but The relative bias  of number of deaths (-0.29) and confirmed cases (-0.04) shows that the overestimation of mortality rate in Italy was duo to under estimation of number of deaths. But in South Korea the relative bias showed that over estimation were the results of over estimation of confirmed cases (0.48) not under estimation of number of deaths (0.14).
Relative ratios showed that predictions in Iran was lower than actual reports (figure 1). Relative bias of number of deaths was remarkably underestimated (0.35) but the number of confirmed cases was almost similar to real ones (-0.03).
The all three models were utilizing the same scenario and the results of predictions Iran was differed. Thus it seems that number of deaths due to Covid-19 in Iran were reported less. The number of predictions of confirmed cases were very acceptable thus the hypothesis of reported deaths from Iran were lower than real ones become stronger.
Shojaee S, Pourhoseingholi MA, Ashtari S, Vahedian-Azimi A, Asadzadeh-Aghdaei H, Zali MR. Predicting the mortality due to Covid-19 by the next month for Italy, Iran and South Korea; a simulation study. Gastroenterology and hepatology from bed to bench. 2020;13(2):177-9.
Neter J, Kutner MH, Nachtsheim CJ, Wasserman W. Applied linear statistical models. 1996.
Rothman AJ, Klein WM, Weinstein ND. Absolute and Relative Biases in Estimations of Personal Risk1. Journal of Applied Social Psychology. 1996;26(14):1213-36.
- Abstract Viewed: 0 times