Non-Invasive Classification of Blood Glucose Level for Early Detection Diabetes Based on Photoplethysmography Signal

Round 1

Reviewer 1 Report

1. The figures in the manuscript are attached under low resolution. It is hard to extract useful information. The quality of the presentation makes it unacceptable for publication. 
2. It is recommended to discuss the principle of measurements in more details. As far as I know, the absorption coefficient of glucose is pretty low in visible and NIR bands. Blood viscosity seems to be a possible bridge but it is a multifactorial assessment, so I am pretty curious about the concept of the measurements conducted here. Though it is not easy to explain the principle of deep learning, a general related discussion is still necessary in methods part.
3. In Figure 8, the author mentioned the evaluation of PPG signal segments. How is the evaluation conducted? Is it manual evaluation or automatic? Is there any standard to classify these signals? What about the individual difference? How to address it?     

Author Response

Dear Editor,

Please see the attachment. 

Regards

Ernia Susana

Author Response File: Author Response.docx

Reviewer 2 Report

Please see the attachment.

Comments for author File: Comments.pdf

Author Response

Dear Editor,

Please see the attachment

Regards

Ernia Susana

Author Response File: Author Response.docx

Reviewer 3 Report

The paper aimed to very important problem, especially for Covid analysis too. The paper is interesting for bioscientists as well as for computer scineces.

However, there are a lot of issues to improve the paper:

1) The figures are presented in low-resolution quality.

2) Equations 1-6 are well-known. They can be skipped.

3) What is the purpose of Fig. 11? The information about ANN is well-known too. As well as the rest of models

4) 2.2.6 - which ensemble do you use (bagging, boosting or stacking?)

5) Lines 725 "Machine learning does not use training data to make any generalizations. In machine learning, there is no explicit training phase or the training phase process is fast. " Can you explain these sentences? I do not agree with it

6) Please add information about the next steps and future plans.

Author Response

Dear Editor,

Please see the attachment ( response to reviewers after an updated manuscript with yellow highlighting indicating changes )

Regards

Ernia Susana

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The author did not fully address my question 2 and 3. A discussion of the measurement principle is very necessary. The pre-processing step still seem to be very abitrary. 

Author Response

Dear Editor,

We are uploading our point-by-point responses to the reviewer, please see the attachment.

Best Regards

Ernia Susana, et al.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors are to be commended for performing such a comprehensive revision of their work. As a result, the current manuscript shows a notably higher quality. I would add just a few minor comments for the authors to keep improving their work:

Were the “acceptable” and “excellent” signals treated the same way?

What are the limitations of the study?

Author Response

Dear Editor,

We are uploading our point-by-point responses to the reviewer, please see the attachment.

Best Regards

Ernia Susana, et al.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors improved the paper. However, a lot of duplicates are presented in the paper (fig 16 and table 4). Well-known information (l460-540) is still given in the paper without explanation why is the reason for presenting well-known facts.

Author Response

Dear Editor,

We are uploading our point-by-point responses to the reviewer, please see the attachment.

Best Regards

Ernia Susana, et al.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

I would like to rephrase my points:

Question 2: The "principle of measurements" I need to see is not the steps and methods the author provided. Even with ML processing, we still cannot learn from nothing. There must be a fundamental pathology about how glucose levels change PPG signals. It can be a prediction, a rough theory or deductive reasoning, but the discussion is needed.

Question 3: To make it clear, let's assume that there are patients in different age groups or with different arterial stiffness performance, can we use the same standard to preprocess the data?   

Author Response

Dear reviewer 1,

We are uploading point-by-point responses to the comments. Please see the attachment.

Best Regards

Ernia Susana, et al

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors took into account comments. Thank you and good luck

Author Response

Dear Reviewer 3,

Thank You for your insightful and valuable suggestions during the review process. It helps us to improve the quality of the manuscript.

Best Regards

Ernia Susana, et al