Predicting of the Coronavirus Disease 2019 (COVID-19) Epidemic Using Estimation of Parameters in the Logistic Growth Model
Round 1
Reviewer 1 Report
The authors did not handle the hold out correctly. This is evident by the prediction intervals. Since they are extrapolating into the future the prediction intervals should get wider (this can be shown mathematically). Furthermore, having done this type of model many many times the "sill" or carrying capacity is very difficult to estimate when there is little data. My guess is that they fit the model to the full data and then used the full data model to predict the hold out sample. This is in correct. Based on what the current work, I do not see any new or novel methodology here. In fact, this whole paper could be recreated in R using the nls() function in just a few lines of code. I would recommend the authors go back and examine the literature and see that this type of model fitting has been done 20 years ago.
Author Response
Response to Reviewer 1 Comments
Point 1: I The authors did not handle the hold out correctly. This is evident by the prediction intervals. Since they are extrapolating into the future the prediction intervals should get wider (this can be shown mathematically). Furthermore, having done this type of model many many times the "sill" or carrying capacity is very difficult to estimate when there is little data. My guess is that they fit the model to the full data and then used the full data model to predict the hold out sample. This is in correct. Based on what the current work, I do not see any new or novel methodology here. In fact, this whole paper could be recreated in R using the nls() function in just a few lines of code. I would recommend the authors go back and examine the literature and see that this type of model fitting has been done 20 years ago.
Response 1: In this study, the prediction interval can be widened, but the spread of Covid-19 is very fast, so we scenario the time interval in 2-4 weeks (1 month or more, so that the severity of this epidemic can be quickly predicted, then the prevention will be carried out by the government of a country quickly too. The carrying capacity is difficult to predict during the early phase of the epidemic, this is because screening and tracking of suspect individuals at the beginning of the phase are not carried out as much as possible so that individuals infected with COVID-19 can be detected sooner so that daily cases and cumulative infected cases will fluctuate. Over time, the more process of screening and tracking for suspect individuals, the epidemic curve pattern can be predicted well by this model. This simple method can still contribute to the spread of COVID-19 in several countries. Although this model does not require a complicated algorithm and can be programmed in any programming language, however, it is easy to apply in today's situation.
Author Response File: 
Author Response.docx
Reviewer 2 Report
After careful examination, I suggest publication of the paper, as I believe the paper is at good standing and the authors revised it properly. There are still some editorial minor errors that I believe can be taken care of during the proofread of the publication process. Please let me know if any additional comments or information is needed.
Author Response
Response to Reviewer 2 Comments
Point 1: After careful examination, I suggest publication of the paper, as I believe the paper is at good standing and the authors revised it properly. There are still some editorial minor errors that I believe can be taken care of during the proofread of the publication process. Please let me know if any additional comments or information is needed.
Response 1: We would like to thank the reviewers for their support of our manuscript to be published, we believe this simple method can contribute to predicting an epidemic in a country with a short prediction with a time interval minimum of 14-28 days (2-4 weeks or 1 month) or more to see the prediction of the growth rate and the cumulative number of infected cases in a final phase in that time interval so the government can take action to prevent the spread of COVID-19 quickly.
We have also revised the grammar and spelling of English.
Author Response File: Author Response.docx
Reviewer 3 Report
Dear Authors,
The paper is well organized. COVID is a recently popular issue.
Please check the spell on the paper.
I think it is worthwhile to publish the paper after minor revisions.
Author Response
Response to Reviewer 3 Comments
Point 1: Dear Authors,
The paper is well organized. COVID is a recently popular issue.
Please check the spell on the paper.
I think it is worthwhile to publish the paper after minor revisions.
Response 1: We would like to thank the reviewers for their support of our manuscript to be published, we believe this simple method can contribute to predicting an epidemic in a country with a short prediction with a time interval minimum of 14-28 days (2-4 weeks or 1 month) or more to see the prediction of the growth rate and the cumulative number of infected cases in a final phase in that time interval so the government can take action to prevent the spread of COVID-19 quickly.
We have also revised the grammar and spelling of English.
Author Response File: Author Response.docx
Reviewer 4 Report
The authors propose data parametrized logistic growth models for predicting the number of infected individuals with COVID-19 within different countries. The methodology is clearly explained and the paper is certainly an interesting contribution for documenting the evolution of the pandemic worldwide. Overall the predictive power of the model may be affected by factors that were not included such as different variants or mutations of the virus. Perhaps considering a logistic model with Allee effect may mitigate the aspect of the variants of the COVID-19 virus which demonstrably have different infectivity rates.
Author Response
Response to Reviewer 4 Comments
Point 1: The authors propose data parametrized logistic growth models for predicting the number of infected individuals with COVID-19 within different countries. The methodology is clearly explained and the paper is certainly an interesting contribution for documenting the evolution of the pandemic worldwide. Overall the predictive power of the model may be affected by factors that were not included such as different variants or mutations of the virus. Perhaps considering a logistic model with the Allee effect may mitigate the aspect of the variants of the COVID-19 virus which demonstrably have different infectivity rates. 

Response 1: In this study, we did not take into account the influence of factors such as different virus variants or mutated viruses in this prediction model, because this prediction model is only based on input data from actual data on cases of COVID-19 infection (for all variants of virus). Because the spread of this virus is very fast, it is necessary to predict for 2 - 4 weeks (1 month) or more to see the growth rate and the cumulative number of COVID-19 infected cases, so that prevention will be carried out by the government of a country immediately. After prevention is carried out, it is necessary to predict again to see the peak time or turning point, as well as to look back at the growth rate and the cumulative number of COVID-19 infected cases.
We have read literature on the Allee effect. At present, this study has not considered a logistics model with the Allee effect that can reduce aspects of the COVID-19 virus variant which is shown to have different levels of infectivity. We will consider a logistic model with the Allee effect for our future research, as this will be an exciting new study, it may be able to become one other research manuscript in the future.
We would like to thank the reviewers for their support of our manuscript to be published, we believe this simple method can contribute to predicting an epidemic in a country with a short prediction with a time interval minimum of 14-28 days (2-4 weeks or 1 month) or more to see the prediction of the growth rate and the cumulative number of infected cases in a final phase in that time interval so the government can take action to prevent the spread of COVID-19 quickly.
We have also revised the grammar and spelling of English
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
While the manuscript has added more countries that have been analyzed the work itself is still problematic. The prediction bands are clearly wrong as they should get wider as you move away from the mean. There is no formula stating how the prediction bands are calculated hence it is difficult to say exactly what is wrong with them. But they do not exhibit the correct behavior. The authors do not use a hold out sample to show the predictive performance (cross-validation) of the model. That is where the true utility of this model is. I would recommend they focus only on two countries who are distinctly different and see how well the model predicts the future. My guess is that it will have a difficult time with the sill (carrying capacity). There are no confidence bands around the inflection point nor the sill (carrying capacity). These are items that are estimated and hence have standard errors associated with them. This should be reflected in the work. The time period is broken up into several subintervals and the model is fit to each. However, there is no statistical test performed to see if the change in the parameters across the subintervals is statistically significant. In nonlinear models R^2 is not a real thing as the variation is not partitioned. They should change it to be the Pseudo-R^2. This is the same quantity that they have already computed however it acknowledges that the sums of squares are not partitioned. The authors should do a complete literature survey to see what is currently being done as this is a very simplistic model. A recent article of interest might be: https://lettersinbiomath.journals.publicknowledgeproject.org/index.php/lib/article/view/319 or https://ir.library.illinoisstate.edu/spora/vol7/iss1/1/ This will give the authors a sense of what models people are using in the current literature.