Diabetic Retinopathy Detection: A Blockchain and African Vulture Optimization Algorithm-Based Deep Learning Framework
Round 1
Reviewer 1 Report
The authors of the proposed methodology have tried to do two things in essence:
1. To enhance the classification performance of DL models in order to detect diabetic retinopathy.
2. To develop a blockchain-based EHR framework that enhances data privacy and security.
Major Comments:
· The goal of this work is vague. What are the main objectives of this work? Is it to enhance the DL methodology or to develop a blockchain-based EHR framework? If the authors want to perform both, what is the motive to present both of these methodologies together? Because, intuitively, a blockchain-based EHR framework does not need to be integrated with a deep-learning pathology detection model for validation.
· What is the state-of-the-art for the blockchain-based EHR frameworks? The authors need to provide an account of the SOTA by performing a comprehensive literature review.
· The motivation is extremely confusing. The authors state, “The illegal attacks detect before exchanging data because of the mutual authentication process in an online healthcare monitoring system”. What do the authors mean by this?
· The authors have explained the AVO algorithm but they have failed to explain the said algorithm works in the context of the extracted features in the proposed work.
Minor Comments:
· The introduction section lacks flow and coherence. For instance, in the second paragraph of the introduction section, the authors begin by talking about EHR security and then suddenly shift the discussion to diabetes.
· The literature review is incomplete and does not present a complete picture of the work done in the domain.
Author Response
Respected Reviewer,
Please find my review comment responses in the attached file and also mentioned below:
Comments:
The authors of the proposed methodology have tried to do two things in essence:
- To enhance the classification performance of DL models in order to detect diabetic retinopathy.
- To develop a blockchain-based EHR framework that enhances data privacy and security.
Major Comments:
- The goal of this work is vague. What are the main objectives of this work? Is it to enhance the DL methodology or to develop a blockchain-based EHR framework? If the authors want to perform both, what is the motive to present both of these methodologies together? Because, intuitively, a blockchain-based EHR framework does not need to be integrated with a deep-learning pathology detection model for validation.
Response: Thank you for the valuable comment. The response to this comment is addressed in the Introduction 3rd paragraph and is highlighted with proper justification and references [7] and [21].
- What is the state-of-the-art for the blockchain-based EHR frameworks? The authors need to provide an account of the SOTA by performing a comprehensive literature review.
Response: Thank you for the input. The comment is addressed and highlighted in the Literature Review section with references [22],[23],[24],[25],[26],[27],[28].
- The motivation is extremely confusing. The authors state, “The illegal attacks detect before exchanging data because of the mutual authentication process in an online healthcare monitoring system”. What do the authors mean by this?
Response: Thank you for the comment. The motivation is revised with the required information and highlighted in section 1.1.
- The authors have explained the AVO algorithm but they have failed to explain the said algorithm works in the context of the extracted features in the proposed work.
Response: Thank you for the comment. The comment and relevance of using AVO algorithm is added in Section 3.2.2
Minor Comments:
- The introduction section lacks flow and coherence. For instance, in the second paragraph of the introduction section, the authors begin by talking about EHR security and then suddenly shift the discussion to diabetes.
Response: Thank you for the input. The Introduction section 2nd paragraph revised ensuring proper work flow is maintained.
- The literature review is incomplete and does not present a complete picture of the work done in the domain.
Response: Thank you for the comment. The comment is addressed in the Literature Review section with references [22],[23],[24],[25],[26],[27],[28].
Author Response File: 
Author Response.docx
Reviewer 2 Report
Summary: Remarks and Suggestions: - Generally, the paper is well-written and well-strucutred and the covered topic is quite interesting. - The authors use two abbreviations in the abstract without defining what they signify (EHR and DR). - The abstract contains some English mistakes: "The results of the classifier is evaluated" (it is essential to perform a proofreading) - The abbreviation TaylorAVOA is proposed as a keyword. However, it is not used in the abstract or in the introduction. In total, it appears only three times in the whole paper (In list of keywords, in the tile of figure 2 and in the conclusion) which is a bit surprising. (Sometimes a different spelling is used: "Taylor-AVO") - Nowhere in the text do the authors define the meaning of the acronym EHR (Only in the title of section 3.1.3). - The writers' contributions are described in the introduction as a list with only one point. The latter may raise concerns regarding the proposed work's quality and the significance of the presented contribution. - It seems to be strange to have a subsection 1.1 but no subsections 1.2 and 1.3, for example.- The authors need to add a short paragraph at the end of the introduction which describes the structure of the paper. - In Section 2 about related works, the studied references are numbered from 20 to 27. However, the references appearing in Table 1 (which is supposed to summarize the section) are numbered from 1 to 8. - Table 1: A column concerning the limitations of the studied works needs to be added. - The authors need to include the following interesting references in their work: + https://www.scitepress.org/PublicationsDetail.aspx?ID=SPtWcseyOGU=&t=1 + https://www.mdpi.com/2227-9032/7/2/56 + https://www.mdpi.com/2227-9032/9/6/712 + https://link.springer.com/article/10.1007/s00779-021-01583-8 + https://www.mdpi.com/1660-4601/19/23/15577 + https://ieeexplore.ieee.org/document/8481969 + https://ieeexplore.ieee.org/document/9432157 - The title of section 3.1 needs to start with a capital letter. - Figure 1 is of poor quality and should be enhanced. - Subsections (3.1.1), (3.1.2), (3.1.3), (3.1.4) and (3.1.5) are components while Subsections (3.1.6), 93.1.7) and (3.1.8) are steps. This makes the paragraph inconsistent - Is there any theoretical way for checking the correctness of Algoritms 1, 2 and 3? - The authors need to provide justification about the adoption of the Taylor-AVO African Vulture Optimization algorithm. - Similarly, more justification about the use of SqeezeNet Structure is needed. - The authors need to identify the limitations of the proposed approach and to propose more future work directions.
- For instance, what is the cost of the adoption of the blockchain technology and its limitations.
Author Response
Dear Reviewer,
Please find the review comments attached in the file and also mentioned below:
Comments:
Summary: Remarks and Suggestions: - Generally, the paper is well-written and well-structured and the covered topic is quite interesting.
- The authors use two abbreviations in the abstract without defining what they signify (EHR and DR). The abstract contains some English mistakes: "The results of the classifier is evaluated" (it is essential to perform proofreading)
Response: Thank you for the comment. The Abstract is revised.
- The abbreviation TaylorAVOA is proposed as a keyword. However, it is not used in the abstract or in the introduction. In total, it appears only three times in the whole paper (In the list of keywords, in the tile of figure 2, and in the conclusion) which is a bit surprising. (Sometimes a different spelling is used: "Taylor-AVO")
Response: Thank you for the comment. The comment is addressed in the introduction section and modified in the entire manuscript.
- Nowhere in the text do the authors define the meaning of the acronym EHR (Only in the title of section 3.1.3). - The writers ‘contributions are described in the introduction as a list with only one point. The latter may raise concerns regarding the proposed work’s quality and the significance of the presented contribution.
Response: Thank you for the comment. The comment is addressed in the Introduction by providing definition of EHR and EHR related papers are reviewed in the Literature review sections.
- It seems to be strange to have subsection 1.1 but no subsections 1.2 and 1.3, for example.
Response: Thank you for the input. Subsection 1.2 is added in the Introduction which provides the organization of the paper.
- The authors need to add a short paragraph at the end of the introduction which describes the structure of the paper.
Response: Thank you for the comment. The subsection Paper Organization is added in the manuscript which describes the structure of the paper.
- In Section 2 about related works, the studied references are numbered from 20 to 27. However, the references appearing in Table 1 (which is supposed to summarize the section) are numbered from 1 to 8.
Response: Thank you for the comment. The References in the Table 1 are modified.
- Table 1: A column concerning the limitations of the studied works needs to be added.
Response: Thank you for the comment. In Table 1 the limitations of each paper are included as a separated column.
- The authors need to include the following interesting references in their work:
+ https://www.scitepress.org/PublicationsDetail.aspx?ID=SPtWcseyOGU=&t=1
+ https://www.mdpi.com/2227-9032/7/2/56
+ https://www.mdpi.com/2227-9032/9/6/712
+ https://link.springer.com/article/10.1007/s00779-021-01583-8
+ https://www.mdpi.com/1660-4601/19/23/15577
+ https://ieeexplore.ieee.org/document/8481969
+ https://ieeexplore.ieee.org/document/9432157
Response: Thank you for the input. The mentioned references are added in the Literature Review Section with references [22],[23],[24],[25],[26],[27],[28].
- The title of section 3.1 needs to start with a capital letter.
Response: Thank you for the comment. The content is revised.
- Figure 1 is of poor quality and should be enhanced.
Response: Thank you for the comment. The quality of the figure is enhanced.
- Subsections (3.1.1), (3.1.2), (3.1.3), (3.1.4) and (3.1.5) are components while Subsections (3.1.6), 93.1.7) and (3.1.8) are steps. This makes the paragraph inconsistent
Response: Thank you for the comment. This comment is addressed wherein the subsections are removed.
- Is there any theoretical way for checking the correctness of Algorithms 1, 2 and 3?
Response: The proper justification of algorithms mentioned in the manuscript followed by the pseudocodes.
Response: Thank you for the comment. The comment is addressed in Section 3.1 wherein the Steps of the algorithm are specified.
- The authors need to provide justification about the adoption of the TaylorAVO (African Vulture Optimization) algorithm.
Response: Thank you for the comment. The justification of using TaylorAVO algorithm is included in Section 3.2.2.
- Similarly, more justification about the use of SqeezeNet Structure is needed.
Response: Thank you for the comment. The justification of using SqueezeNet structure is included in Section 3.2.3.
- The authors need to identify the limitations of the proposed approach and to propose more future work directions.
Response: Thank you for the comment. The limitations and future work are mentioned and highlighted in the Conclusion Section 5.
- For instance, what is the cost of the adoption of the blockchain technology and its limitations.
Response: Thank you for the comment. The cost of adoption of blockchain technology and the related economic challenges are included in the conclusion Section 5.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
1. The authors must revise algorithm 3. They have provided a pseudocode of AVO which is basically copied from the original AVO publication. There is no need to provide the pseudocode of AVO because it is a well-known algorithm. The authors should instead present a pseudocode of the TaylorAVO algorithm, which is the contribution of this work.
2. The authors need to present the description of the data in light of the robust ML benchmarks as per the following publication:
Mincu, D., Roy, S. Developing robust benchmarks for driving forward AI innovation in healthcare. Nat Mach Intell 4, 916–921 (2022). https://doi.org/10.1038/s42256-022-00559-4.
3. The visibility of the bar graphs in the results section needs to be improved.
Author Response
Reviewer 3：
Summary: Remarks and Suggestions:
- The authors must revise algorithm 3. They have provided the pseudocode of AVO which is basically copied from the original AVO publication. There is no need to provide the pseudocode of AVO because it is a well-known algorithm. The authors should instead present a pseudocode of the TaylorAVO algorithm, which is the contribution of this work.
Response: Thank you for the comment. The comment is addressed in Section 3.2.2 wherein Equation 5 and related explanation is added. Also, Algorithm 3 is updated.
- The authors need to present the description of the data in light of the robust ML benchmarks as per the following publication:
Mincu, D., Roy, S. Developing robust benchmarks for driving forward AI innovation in healthcare. Nat Mach Intell 4, 916–921(2022). https://doi.org/10.1038/s42256-022-00559-4.
Response: Thank you for the comment. The comment is addressed in Section 4.1 Dataset Description.
- The visibility of the bar graphs in the results section needs to be improved.
Response: Thank you for the comment. The comment is addressed for all bar graphs and visibility is improved.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors considered all my comments. I have no other suggestion. Good luck.
Author Response
Dear Reviewer,
Thank you for your valuable comments and positive consideration of the previously addressed comments.
Thank you
Round 3
Reviewer 1 Report
Thank you for addressing all my comments.
