Value-Added Carp Products: Multi-Class Evaluation of Crisp Grass Carp by Machine Learning-Based Analysis of Blood Indexes

Round 1

Reviewer 1 Report

This is a very innovative article showing the possibilities of using machine learning-based analysis of blood indexes to evaluate value-added carp products. These techniques are increasingly used to analyze food parameters due to the precise results.
Comments to the text of the article concern mainly the editorial side.

1. Introduction
The content on lines 69-79 clearly indicates a summary that should not be included here. I would expect a clear formulation of the purpose of this research, as I am asking the Authors.

2. Methodology - sensory evaluation
Has anyone described how to perform the sensory evaluation of value-added carp products? Is this a standard practice in such studies? Or maybe it was idea of the Authors - who defined such a procedure during the sensory evaluation of products? Please explain and indicate the literature, if it was used by the Authors.
3. Conclusions
The content on lines 322-325 should not be included in the "Conclusions" chapter as it indicates the limitations of the method used. Meanwhile, the conclusions should give a clear signal related to the results obtained in the study.
4. Literature
Please pay attention to the correctness of the quotation of literature items - there are minor errors in the record of items 5, 11, 17 and 20.  

Author Response

Reviewer 1:

This is a very innovative article showing the possibilities of using machine learning-based analysis of blood indexes to evaluate value-added carp products. These techniques are increasingly used to analyze food parameters due to the precise results.

Comments to the text of the article concern mainly the editorial side.

Thank you for your kind and constructive comments on our manuscript. According to your comments, we have revised the manuscript as follows. In addition to the changes replying to reviewer's comments, we corrected some English expressions. All changes are indicated in the revised manuscript.

1. Introduction

The content on lines 69-79 clearly indicates a summary that should not be included here. I would expect a clear formulation of the purpose of this research, as I am asking the Authors.

Reply:

As suggested, we revised this paragraph to include the purpose of this study.

2. Methodology - sensory evaluation

Has anyone described how to perform the sensory evaluation of value-added carp products? Is this a standard practice in such studies? Or maybe it was idea of the Authors - who defined such a procedure during the sensory evaluation of products? Please explain and indicate the literature, if it was used by the Authors.

Reply:

We used the sensory evaluation protocol that has been used for this grass carp product for a long time. Details are described in Yang et al (2015). We cited this paper in the section 2.2.

3. Conclusions

The content on lines 322-325 should not be included in the "Conclusions" chapter as it indicates the limitations of the method used. Meanwhile, the conclusions should give a clear signal related to the results obtained in the study.

Reply:

According to this comment, we moved the limitation part to the end of Results and Discussion. We also modified some sentences, and the conclusions in the revised version is now based on the results obtained in this study.

4. Literature

Please pay attention to the correctness of the quotation of literature items - there are minor errors in the record of items 5, 11, 17 and 20.

Reply:

We corrected items 5, 17, and 20 according to this comment. We double-checked the reference 11 but could not find any errors in this citation. We would be grateful if you can point out the error, if any, in the next revision when necessary.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript is interesting scientific contributions to study the value-added carp products: multi-class evaluation of crisp grass carp by machine learning-based analysis of blood indexes. In this regard, the aim of this work was to assess quality (muscle hardness) of crisp grass carp by combination of machine learning and blood analysis. In addition, seven indexes including hydrogen peroxide (H2O2), G6PD, reduced glutathione (GSH), malondialdehyde (MDA), red blood cells (RBC), platelet count (PLT) and lymphocytes (LY) were also analyzed. The paper has high scientific level, the experiment is well designed, the discussion is consistent and the final conclusions are interesting. Therefore, the manuscript need major revision.

Suggestions for edition as well as some comments are the following:

Abstract

Please, add numerical results in the Abstract section to support the main findings cited.

Keywords

Please change these keywords “crisp grass carp; machine learning” because they are presented in the title.

Material and methods

Sensory evaluation. The number of panellist is very low (only 5 trained experts).

Results

Please show all obtained results in tables

Hardness, please express the results in Newton

I did not see the results of sensorial analysis in the manuscript.

Author Response

Reviewer 2:

The manuscript is interesting scientific contributions to study the value-added carp products: multi-class evaluation of crisp grass carp by machine learning-based analysis of blood indexes. In this regard, the aim of this work was to assess quality (muscle hardness) of crisp grass carp by combination of machine learning and blood analysis. In addition, seven indexes including hydrogen peroxide (H2O2), G6PD, reduced glutathione (GSH), malondialdehyde (MDA), red blood cells (RBC), platelet count (PLT) and lymphocytes (LY) were also analyzed. The paper has high scientific level, the experiment is well designed, the discussion is consistent and the final conclusions are interesting. Therefore, the manuscript need major revision.

Suggestions for edition as well as some comments are the following:

Thank you for your kind and constructive comments on our manuscript. According to your comments, we have revised the manuscript as follows. In addition to the changes replying to reviewer's comments, we corrected some English expressions. All changes are indicated in the revised manuscript.

Abstract

Please, add numerical results in the Abstract section to support the main findings cited.

Reply:

The word limitation has made this revision a little difficult, but we included numerical results as much as possible in the result part of Abstract (L26-L29).

Keywords

Please change these keywords “crisp grass carp; machine learning” because they are presented in the title.

Reply:

We removed these keywords as suggested and added "muscle hardness" instead. According to this change, "muscle quality" was changed to "meat quality."

Material and methods

Sensory evaluation. The number of panellist is very low (only 5 trained experts).

Reply:

The five panelists were experienced experts who have been participated in sensory evaluation of crisp grass carp products for a long time. We added a citation describing this in section 2.2 (Reference 4, Yang et al., 2015; another reviewer also suggested to add the citation), in which we performed a detailed analysis to compensate for the weakness of the sensory evaluation. In Yang et al. (2015), results obtained from five experts were consistent and highly correlated with those of instrumental evaluation methods, and the present study used the verified method. In addition to this, sensory evaluation results were consistent with those of boxplot analysis in this study (Figure 1). Together, we consider that the sensory evaluation results were reliable enough to draw the conclusion. We also modified the second paragraph of the section 3.1. to describe the above points.

Results

Please show all obtained results in tables

Reply:

All obtained results were already added in Supplementary files 1-4. Supplementary file 1: Raw Data of muscle hardness from ordinary grass carp and crisp grass carp; Supplementary file 2: Raw Data of muscle hardness from ordinary grass carp and crisp grass carp; Supplementary file 3: Raw Data of seven blood indexes from ordinary grass carp and crisp grass carp; Supplementary file 4: Parameters of six machine learning models.

Comment:

Hardness, please express the results in Newton

Reply:

Corrected.

Comment:

I did not see the results of sensorial analysis in the manuscript.

Reply:

We added the results as the second paragraph of the section 3.1.

Author Response File: Author Response.docx

Reviewer 3 Report

The title and purpose of the research clearly describe the scope of the work done. The overall quality of the manuscript is good, very good English, the experience was properly designed, all information to understand the proposes of the manuscript are correctly presented. The research material has been properly prepared, the number of samples, analytical techniques and statistical methods used are adequate for this type of experience, they have been accurately and clearly described in the manuscript of the paper. The results have been collected in tables, and graphs and are correctly presented (except for texture parameters units). I believe that the work submitted to me for evaluation is very valuable in providing a lot of new scientific information on the use of a new technique to predict the quality of fish meat. Since the manuscript shows to be a very good developed job. My recommendation is that the manuscript need minor revision.

Minor Comments:

Line 36. Include an space between “C.” and “et”

Line 55 and line 78. Put “in vivo” in italics.

Line 57. Delete the information between brackets “[e.g. random forest…etc..].” It is unnecessary.

Line 59. Please, if you delete the information between brackets in line 57, define “SVM”: “support vector machine (SVM)”.

Line 66. Define “RF” as “random forest (RF)”.

Line 73. Please, define “G6PD”.

Line 75. Please, define “NB”.

Line 81-93. As your original descriptions of research conducted in experimental animals contained in your manuscript, you should provide the details of approval by a properly constituted research ethics committee. As a minimum, the project identification code and name of the ethics committee or institutional review board should be cited in the Methods section.

Line 91-93. I have to say that it is not my field of knowledge, so I consider this part very confusing and I do not understand it. I consider, from my point of view, that it is not fully explained, so it is not clear. How much and what approximate dimensions of muscle was removed from each animal? Were the same animals always used or were the animals slaughtered after extraction? The sampling as well as the size of the extracted muscle should be much better explained. Very little information is given, and they are really confusing. A total of 540 animals are bred, but only 252 animals are selected for the study. Please clarify all this part of the material and methods, being very clear how the muscles have been sampled, and if they were always (30, 60, 90 and 120d) the same animals or different. Please, also explain the sampling procedure.

Line 98.  “mm·s-1” Put “-1” in superscript.

Line 191. “TP” appears twice in the denominator of the equation. I think it is wrong; authors must remove one of them. Please check this out.

Figure 1. Why the authors use “g” as hardness unit?. The international system says that the texture (hardness) should be expressed in newtons (N), despite several authors use “g” or “kg”. Please, explain correctly why you use these units or, in my opinion the best option, change the “g” per “N”, and correct the manuscript according the new data.

Figure 2. In the figure title, please put the “2” of H2O2 in subscript.

Table 1. Authors should ensure that the tables are not cut between different pages. To prevent part of the information related to the same machine learning mode from being cut between two different pages, they must make a table for each machine learning model.

Author Response

Reviewer 3:

The title and purpose of the research clearly describe the scope of the work done. The overall quality of the manuscript is good, very good English, the experience was properly designed, all information to understand the proposes of the manuscript are correctly presented. The research material has been properly prepared, the number of samples, analytical techniques and statistical methods used are adequate for this type of experience, they have been accurately and clearly described in the manuscript of the paper. The results have been collected in tables, and graphs and are correctly presented (except for texture parameters units). I believe that the work submitted to me for evaluation is very valuable in providing a lot of new scientific information on the use of a new technique to predict the quality of fish meat. Since the manuscript shows to be a very good developed job. My recommendation is that the manuscript need minor revision.

Thank you for your kind and constructive comments on our manuscript. According to your comments, we have revised the manuscript as follows. In addition to the changes replying to reviewer's comments, we corrected some English expressions. All changes are indicated in the revised manuscript.

Comment:

Line 36. Include an space between “C.” and “et”

Line 55 and line 78. Put “in vivo” in italics.

Line 57. Delete the information between brackets “[e.g. random forest…etc..].” It is unnecessary.

Line 59. Please, if you delete the information between brackets in line 57, define “SVM”: “support vector machine (SVM)”.

Line 66. Define “RF” as “random forest (RF)”.

Reply:

All corrected accordingly.

Comment:

Line 73. Please, define “G6PD”.

Line 75. Please, define “NB”.

Reply:

This part was deleted according to a comment from another reviewer. We instead defined these words in sections 2.4. and 2.5., respectively.

Comment:

Line 81-93. As your original descriptions of research conducted in experimental animals contained in your manuscript, you should provide the details of approval by a properly constituted research ethics committee. As a minimum, the project identification code and name of the ethics committee or institutional review board should be cited in the Methods section.

Reply:

We added the approval at the end of the section 2.1.

Comment:

Line 91-93. I have to say that it is not my field of knowledge, so I consider this part very confusing and I do not understand it. I consider, from my point of view, that it is not fully explained, so it is not clear. How much and what approximate dimensions of muscle was removed from each animal? Were the same animals always used or were the animals slaughtered after extraction? The sampling as well as the size of the extracted muscle should be much better explained. Very little information is given, and they are really confusing. A total of 540 animals are bred, but only 252 animals are selected for the study. Please clarify all this part of the material and methods, being very clear how the muscles have been sampled, and if they were always (30, 60, 90 and 120d) the same animals or different. Please, also explain the sampling procedure.

Reply:

We revised the section 2.1. to include details of sampling. All samples were different – namely, there was no repeated sampling, and one individual produced only one data. The size of white muscle sample was also included, and we hope that the section is now clearer.

Comment:

Line 98.  “mm·s-1” Put “-1” in superscript.

Line 191. “TP” appears twice in the denominator of the equation. I think it is wrong; authors must remove one of them. Please check this out.

Reply:

Thank you for pointing them out, especially for the second one. We corrected the mistake.

Comment:

Figure 1. Why the authors use “g” as hardness unit?. The international system says that the texture (hardness) should be expressed in newtons (N), despite several authors use “g” or “kg”. Please, explain correctly why you use these units or, in my opinion the best option, change the “g” per “N”, and correct the manuscript according the new data.

Figure 2. In the figure title, please put the “2” of H2O2 in subscript.

Reply:

We corrected the unit of Figures. 1 and 2 accordingly.

Comment:

Table 1. Authors should ensure that the tables are not cut between different pages. To prevent part of the information related to the same machine learning mode from being cut between two different pages, they must make a table for each machine learning model.

Reply:

According to this comment, we split the table into Table 1 and Table 2. Each table contains three machine learning models. This is a little different from the reviewer's suggestion, but now the information of the same model can be seen in one page. We hope that this change would be acceptable.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The manuscript was greatly improved.