Supervised Learning by Evolutionary Computation Tuning: An Application to Blockchain-Based Pharmaceutical Supply Chain Cost Model
Round 1
Reviewer 1 Report
Dear Authors
This paper analyzes a pharmaceutical supply chain (PSC) using a cost-minimization mathematical model with blockchain technology (BT). It aims to improve the safety, performance, and transparency of medical information sharing in a healthcare industry.
This paper shows a good fit with the journal ‘Mathematics’ to which it was submitted and performed quite well in mathematical modeling analysis. However, overall, there is a lack of review of previous studies, and there is a significant lack of description of the differences between the paper and previous studies. In addition, the key contributions, values, implications and insights of this paper are missing. Accordingly, my decision is 'major revision', and the authors are requested to make every effort to revise and supplement the manuscript.
(1) In the introduction, please describe in more detail the differences from previous studies and the excellence of this study.
More specifically, what are the knowledge gap in this stream of literature this study intends to fill up? Please clarify the novelty of this paper and new knowledge authors want to provide more in detail.
(2) Please describe the research questions in this work and add background interesting story telling about the pharmaceutical supply chain and research motivation to the introduction. Also, please add relevant previous studies on the pharmaceutical supply chain to Section 2.
More particularly, when I searched for ‘Blockchain Technology Pharmaceutical Supply Chain’ on the Web of Science, I was able to find a total of 93 papers. Among them, there are a total of 14 review papers. Again, among these 14 review papers, please pay attention to the following five papers particularly.
- Ghadge, A., Bourlakis, M., Kamble, S., & Seuring, S. (2022). Blockchain implementation in pharmaceutical supply chains: A review and conceptual framework. International Journal of Production Research, 1-19.
- Bamakan, S. M. H., Moghaddam, S. G., & Manshadi, S. D. (2021). Blockchain-enabled pharmaceutical cold chain: Applications, key challenges, and future trends. Journal of Cleaner Production, 302, 127021.
- Saeed, G., Kohler, J. C., Cuomo, R. E., & Mackey, T. K. (2022). A systematic review of digital technology and innovation and its potential to address anti-corruption, transparency, and accountability in the pharmaceutical supply chain. Expert Opinion on Drug Safety, 21(8), 1061-1088.
- Elangovan, D., Long, C. S., Bakrin, F. S., Tan, C. S., Goh, K. W., Yeoh, S. F., ... & Ming, L. C. (2022). The use of blockchain technology in the health care sector: systematic review. JMIR medical informatics, 10(1), e17278.
- Sharma, A., Kaur, S., & Singh, M. (2021). A comprehensive review on blockchain and Internet of Things in healthcare. Transactions on Emerging Telecommunications Technologies, 32(10), e4333.
(3) In the Conclusion and Discussion section, please add the implications and insights of the study.
(4) In section 4, please add a more specific description of the assumptions of the mathematical model in the context of the pharmaceutical supply chain.
Authors should derive and support the assumptions of these mathematical models theoretically and practically based on the previous studies in the literature and/or actual processes in real organizations.
(5) Please explain the research methodology and data learned in more detail.
(6) Numerical values appearing in tables and some mathematical symbols are very difficult to read.
Also, it has not been edited according to the journal editorial guidelines. So, please proofread and edit this paper completely.
(7) Details such as calculation time and computing environment with the algorithm are missing.
(8) The parameters, variables, and constraints of the mathematical model proposed in Table 1 are mixed. Authors should categorize them into groups and typeset them so that readers can easily distinguish which symbols correspond to parameters, variables, and constraints respectively.
In addition, some mathematical symbols are not italicized and the mathematical notations and equations are not typeset professionally very well.
Author Response
Dear Reviewer 1,
Thank you so much for your comment, which improve our papers.
Regards,
Authors
Comment 0: There is a lack of review of previous studies.
Response: The paper contains a literature review with this sub-heading: 2.1. PSC and its components for a hospital; 2.2. BT drives PSC; 2.3. Evolutionary Computation and Supervised Learning; 2.3.1. ACO and FA; 2.3.2. KNN, DT, NB, and SVM. These parts provide a comprehensive background for the domain of BT-enabled PSC. Thank you.
Comment 1: In the introduction, please describe in more detail the differences from previous studies and the excellence of this study. what are the knowledge gap in this stream of literature this study intends to fill up? Please clarify the novelty of this paper and new knowledge authors want to provide more in detail.
Response: We explain it in some parts. For instance, lines 50, 58, 59 describe the difference between the paper and previous studies. Generally, this is a new cost mathematical model for the BT-enabled PSC system.
The knowledge gap is in the line 51 when the paper says, “the present study seeks to address the cost problem of the BT-based PSC.”
The novelty of this paper is to provide the cost mathematical model for BT-enabled PSC system, is to provide a PSC system with BT, and the combination of EC and SL algorithms. All state in lines 50, 59, 62, and 74.
Comment 2: Please describe the research questions in this work and add background interesting story telling about the pharmaceutical supply chain and research motivation to the introduction. Also, please add relevant previous studies on the pharmaceutical supply chain to Section 2. More particularly, when I searched for ‘Blockchain Technology Pharmaceutical Supply Chain’ on the Web of Science, I was able to find a total of 93 papers. Among them, there are a total of 14 review papers. Again, among these 14 review papers, please pay attention to the following five papers particularly.
Response: We selected three general research questions: (i) What are the cost components of the BT-based PSC cost model in a hospital, and what is the mathematical cost model? (ii) Which algorithms show better performance in minimising the prediction errors of the BT-based PSC cost model? (iii) What are the important cost components of the model? I understand that a paper should support its hypotheses with relevant literature, but we do not have hypotheses. A question like What are the cost components of the BT-based PSC cost model has no background and it is a basic question. But we try to respond these questions in this paper. In the Discussion section (6), we respond to them one by one. We also added some sentences in the 4. Mathematical cost model for BT-based PSC section to make it clearer. Thank you.
Thank you for introducing some references. Section 2 contains relevant references about BT in PSC especially in 2.1. PSC and its components for a hospital, 2.2. BT drives PSC, and 2.3. Evolutionary Computation and Supervised Learning. In addition, we already use the reference you suggested entitled “Blockchain-enabled pharmaceutical cold chain: Applications, key challenges, and future trends”. Thank you.
Comment 3: In the Conclusion and Discussion section, please add the implications and insights of the study.
Response: The structure of our study is based on these sources:
1- https://www.nature.com/scitable/topicpage/scientific-papers-13815490/
2- English Communication for Scientists (Unit 2: Writing Scientific Papers) https://www.nature.com/scitable/ebooks/english-communication-for-scientists-14053993/118519636/
The discussion of our study includes:
1- Discuss what the results mean in order of most to least important.
2- Compare the results with other studies.
3- Respond to the original question or objectives outlined in the Introduction section. We have three research questions that will be mentioned in the Introduction section: (i) What are the cost components of the BT-based PSC cost model in a hospital, and what is the mathematical model? (ii) What algorithms behave better in minimizing the prediction errors of the BT-based PSC cost model? (iii) What are the important cost components of the model?
4- Describe the study’s limitations to show reviewers and readers that we have considered our experiment’s weaknesses.
5- End the discussion with a short summary or conclusion regarding the significance of the work. Don't leave the reader thinking "So what?"
6- Include future research directions in your discussion.
In the conclusion, we follow this direction:
1- Main conclusions
2- The most important outcome of your work related to the motivation mentioned in the introduction
3- We have succeeded in addressing the need stated in the Introduction.
4- The short explanation of the topic and then the purpose of the research.
5- This is a short Conclusion as we have a rich discussion section
Comment 4: In section 4, please add a more specific description of the assumptions of the mathematical model in the context of the pharmaceutical supply chain. Authors should derive and support the assumptions of these mathematical models theoretically and practically based on the previous studies in the literature and/or actual processes in real organizations.
Response: We think that the assumptions are clear enough. For example, “The planning horizon is one year” or “The drug expiration date is constant” or “At the beginning of the planning horizon, the age and the number of medications available in the hospital (including the initial stock) are zero” etc. have clear explanation in the pharma domain. They are relevant to medicine and drugs.
The assumptions drive from the references mentioned in the beginning of section 4. We follow the assumptions but matched them with our model.
We revised some sentences at the beginning of section 4 to make it more clear according to your comment. Thank you.
Comment 5: Please explain the research methodology and data learned in more detail.
Response: Section 3, Methodology and Data Generation, completely explain our methodology with a flowchart (figure 1). Lines 237 to 246 also explain data generation. Data is uploaded to a website (line 246). I also attached the code for Python used to generate data.
Comment 6: Numerical values appearing in tables and some mathematical symbols are very difficult to read.
Response: I inform the journal to see if we can make it bigger. Thank you.
Comment 7: Details such as calculation time and computing environment with the algorithm are missing. We cannot use this info in the results because we use TRS to select the better algorithms according to the performance metrics.
Response: In this research, the authors have decided to use four performance metrics to evaluate the algorithms (MSE, RMSE, MAE, and R2), and the time factor is not included in this selection. Table 8 illustrates the ranking of eight selected algorithms based on TRS scores through four performance metrics.
Comment 8: The parameters, variables, and constraints of the mathematical model proposed in Table 1 are mixed. Authors should categorize them into groups and typeset them so that readers can easily distinguish which symbols correspond to parameters, variables, and constraints respectively.
Response: Table 1 was organized and revised according to your comment. Thank you.
Thank you so much for your comments.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents anew PSC mathematical cost model, including blockchain technology, that can improve the safety, performance and transparency of medical information sharing in the healthcare system.
The abstract can be improved by presenting at first the global context, the problem that will be solved in the paper and how it will be solved. Then, it could be nice to illustrate concepts and methods and presents perspectives.
Please avoid the term “we” in the text.
Please try to suppress one “follows” (line 72 and line 83).
Good introduction!!
For the section 2, it could be nice to start by presenting the sub-sections before focusing on each sub-section.
Line 116: Please add the reference number!!!
Line 123: A sentence is required to synthetize the sub-section and create the link with the following.
At the beginning of the sub-section 2.2 a link has to be done with the previous sub-section.
Line 134: add reference number. Please do the same in all the text.
A synthesis is required at the end of the sub-section 2.2 and a link at the beginning of the sub-section 2.3.
Line 228: there is a mistake.
Please add a synthesis of the section 2 and introduce the next section.
Please pay attention on repetition of words such as “Then” in line 272, 274, and 278!!!
A synthesis is required at the end of the section 3 for synthetizing the section and introducing the next section.
Line 423: space between 34 and 27.
A logical link is required before the section 4 and the section 5 for justifying each step of result. This important to understand the logic of what has been done in the part 5.
The conclusion art can be completed by perspectives!
Good paper!
Comments for author File: Comments.pdf
Author Response
Dear Reviewer 2,
Thank you so much for your comment, which improve our papers.
Regards,
Authors
Reviewer 2:
Comment 1: The abstract can be improved by presenting at first the global context, the problem that will be solved in the paper and how it will be solved. Then, it could be nice to illustrate concepts and methods and presents perspectives.
Response: We stated all parts including the global context, the problem, and methods. Please see the following screen. Thank you.
Comment 2: Please avoid the term “we” in the text.
Response: According to the direction in some publisher, we is acceptable, but he-she-I should be avoided. Please see this link as an example: https://www.nature.com/scitable/topicpage/scientific-papers-13815490/ .
Comment 3: Please try to suppress one “follows” (line 72 and line 83).
Response: We keep the word in line 83 and changed it in line 72. Thank you.
Comment 4: For the section 2, it could be nice to start by presenting the sub-sections before focusing on each sub-section.
Response: It was added. Thank you.
Comment 5: Line 116: Please add the reference number!!!
Response: We checked all the manuscript to add the reference number.
Comment 6: Line 123: A sentence is required to synthetize the sub-section and create the link with the following. At the beginning of the sub-section 2.2 a link has to be done with the previous sub-section.
Response: We added it. Thank you.
Comment 7: Line 134: add reference number. Please do the same in all the text.
Response: We checked all the manuscript to add the reference number.
Comment 8: A synthesis is required at the end of the sub-section 2.2 and a link at the beginning of the sub-section 2.3.
Response: We added it. Thank you.
Comment 9: Line 228: there is a mistake.
Response: It is correct now. Thanks.
Comment 9: Please add a synthesis of the section 2 and introduce the next section.
Response: It was added.
Comment 10: Please pay attention on repetition of words such as “Then” in line 272, 274, and 278!!!
Response: We changed 274 and 178 and kept 272. Thanks.
Comment 11: A synthesis is required at the end of the section 3 for synthetizing the section and introducing the next section.
Response: We added a sentence. Thanks.
Comment 12: Line 423: space between 34 and 27.
Response: It was done.
Comment 13: A logical link is required before the section 4 and the section 5 for justifying each step of result. This important to understand the logic of what has been done in the part 5.
Response: It was added.
Comment 14: The conclusion art can be completed by perspectives!
Response: I think there are some sentences in the conclusion to state the new aspect of the study. Please see this sentence in this section and the following sentences: “This study is important because it provides a PSC system with BT (BT Transaction cost and BT Installation cost) that can improve the safety, performance, and transparency of medical information sharing in a healthcare system.” (Line 625 and 627-628).
Comment 16: How to estimate the cost of the BT-based PSC model?
Response: Four SL algorithms predict the costs. We state it several time in the paper (Like line 272-273 in them methodology section).
Comment 17: How to select algorithms with the minimum prediction errors?
Response: Your question has two different parts. We added this sentence in section 3 to response to algorithm selection: These algorithms are well-known and can be successfully applied to solve many engineering problems, facilitating the discussion of their behaviours in our new cost model. All of these algorithms are famous.
Regarding the minimum prediction errors, we added this sentence: The parameters of SL algorithms are usually set empirically, and it takes much time to test and find the best predictive performance of the model. Therefore, the EC algorithm explores possible combinations of parameters, optimize hyperparameters of the SL algorithms, and reduce the prediction errors of the SL algorithms.
Comment 18: How to determine the cost components of the model?
Response: Thank you for this comment. We revised the first paragraph the section 4 to cover this point.
Comment 19: How the different algorithms are created, stored, extracted? How they are adapted to the problem to solve and how the comparison is done?
Response: Back to comment 16, these are the famous algorithms in SL and EC. We do not create an algorithm in the research. We first model the cost of BT-based PSC system, then predict the cost of the model with SL. In this procedure, we use EC to optimize the hyperparameters of the SL. Regarding comparison, we used four performance metrics (MSE, RMSE, MAE, and R2) and TRS method to select the better algorithms. Tables 2, 3, 5, 6, and 8 and Figure 3 show the comparison.
Comment 20:
Response: We generally explain random module applied randint() to generate data in lines 237 to 242. There is a code in Python and not formula. I attached the code called: Data generation Code.
Combination of the EC and SL is also a code in MATLAB. I attached two codes called: ACO - SL - MATLAB Code and FA - SL - MATLAB Code.
Comment 21:
Response: Comment 16 and 19 responded to this comment.
Comment 22:
Response: In the mentioned lines, we clearly explain it. In comment 19, you can check the two attached codes. Figure1, Figure 4, Table 2, 4, 5, 7, 9 are presented the impact of the FW on the model.
Comment 23:
Response: Please look at line 271 and Figure 1.
Comment 24:
Response: Section 3 is related to the mathematical model, and the previous responses help you to point out more about the combination of EC and SL.
Comment 25:
Response: The calculation of Total Ranking Score is explained in line 279 to 283 as you mentioned. Table 8 and Table 9 are exactly presented TRS in detail. Figure 3 and Figure 4 are also described TRS.
Comment 26:
Response: We try to accept your comments and make other more clear. So, it seems there is no need to change Figure 1. Thank you.
Comment 27:
Response: It is done.
Thank you so much.
Author Response File: Author Response.pdf
Reviewer 3 Report
A very well prepared scientific text.
The research background is provided, the main research assumptions are described in a very clear and lucid manner. The aims of the study are indicated - they could be made more explicit, together with a thesis/hypothesis relating to the general understanding of the topic, not limited to the research aspect only.
Methodology clearly described, based on relevant literature.
The study is conducted in a scientific manner, using and selecting an appropriate scientific and research background. The methodological scheme was correct, the study itself was carried out without scientific flaws.
A very good, interesting and necessary text, with both scientific and practical relevance.
Figure 2 is unreadable in normal format, perhaps it could be reformatted.
Author Response
Dear Reviewer 3,
Thank you so much for your comment, which improve our papers.
Regards,
Authors.
Reviewer 3:
- Figure 2 is unreadable in normal format, perhaps it could be reformatted.
Response: Sure. We attached the high-quality figure. Thank you.
Author Response File: 
Author Response.docx
Round 2
Reviewer 1 Report
(1) In the introduction, the authors hope to more clearly highlight the differences between this study and previous studies. Also, please describe the excellence of this study, especially in terms of mathematical modeling.
(2) In the introduction, please add more industry storytelling that suggests the significance of this study. By doing so, I think more readers will find this study interesting.
(3) The authors argue that the assumptions introduced in this study in section 4 are valid. In this regard, please add a description of the situation in the hospitals and health industry in more detail and add references.
(4) Authors are requested to provide details of the numerical experimental conditions and calculation times tested in this study. And can one or two experiments tell us that the results of our analysis are valid? Authors should be able to justify their analyzes with more thorough numerical experiments.
(5) Authors are requested to provide a more thorough discussion than a simple description by providing logical grounds while basing the analysis results on previous studies.
Author Response
Dear Reviewer 1,
Thank you for your helpful comments.
Regards,
Authors.
(1) In the introduction, the authors hope to more clearly highlight the differences between this study and previous studies. Also, please describe the excellence of this study, especially in terms of mathematical modeling.
Response: Thank you for this point. I added some notes to the second paragraph of the introduction section.
(2) In the introduction, please add more industry storytelling that suggests the significance of this study. By doing so, I think more readers will find this study interesting.
Response: Thanks. We added a short story at the end of the first paragraph in the introduction section.
(3) The authors argue that the assumptions introduced in this study in section 4 are valid. In this regard, please add a description of the situation in the hospitals and health industry in more detail and add references.
Response: We added one paragraph to explain the situation and the assumptions in more details based on two references.
(4) Authors are requested to provide details of the numerical experimental conditions and calculation times tested in this study. And can one or two experiments tell us that the results of our analysis are valid? Authors should be able to justify their analyzes with more thorough numerical experiments.
Response: Thank you. We explain it clearly in the first paragraph of the section 5 Results. These sentences were added: “The authors run each algorithm (FA-KNN, FA-DT, FA-SVM, FA-NB, ACO-KNN, ACO-DT, ACO-SVM, and ACO-NB) for 10 runs (totally eighty runs) with 1000 iterations. These experiments help us to be more valid.“
(5) Authors are requested to provide a more thorough discussion than a simple description by providing logical grounds while basing the analysis results on previous studies.
Response: The discussion section explains the results of the study based on the output data. As we mentioned in the paper, we determined the related cost components to design our model and then generated raw data. Therefore, it is not possible to discuss the result based on the results of the other studies. Thanks to you for sending us some papers in this field. Those papers also do not include a cost mathematical model like ours, so how can we compare studies that do not have a cost model with this study? We have a new paper on our hands in this field and should follow your advice because both papers have cost models, and we can compare them. Thank you.
End.
Author Response File: Author Response.docx
