Natural Computing-Based Designing of Hybrid UHMWPE Composites for Orthopedic Implants

Round 1

Reviewer 1 Report

The paper studies the use of Artificial Neural Network (ANN) in combination with Genetic Algorithm (GA) to find optimal combinations of reinforcements to have ultra-high molecular weight polyethylene (UHMWPE) composites with superior mechanical and tribological properties. The studied topic is exciting, and the experiment results seem interesting. However, the whole paper is let down by poor writing and presentation. It can only be reconsidered after a significant revision. The following are some points to consider to improve the paper:

1) The introduction is not clear. The problem is adequately described but current state-of-the-art approaches, what is proposed, and its novelty are all in a blur. The introduction is quite long but has only two paragraphs. One of them consists of 78 lines. The other is also 30 lines. The author might want to break them up into much smaller paragraphs and make the idea clear in each one of them.
2) The descriptions of methods are also not clear or not adequate. GA   is only described by words without a main reference. It is unclear what are its inputs and outputs and how it is connected to the ANN models (what outputs of ANN are used as the input of GA). Nowadays, DNN is much more popular. Even when people use the term ANN they might still use several hidden layers. So please make it clear that you are using only one hidden layer. Also, I believe you have 5 ANN models because you have one model for each output. It was not clear from my first time reading the paper.   
3) The authors never discuss what software/packages are used, what parameters are selected, or the configuration of the machine to run the experiments on.  

There are many small typos or unusual choices of words. Here are some suggestions for changes:
l.19: the optimum -> an optimal
l.32 : has ->have
l.38: two year latter -> the last two years
l.38: because of -> due to
l. 42: thereof -> therefore?
l. 60: and indeed?
l.69: with -> to
l. 95: enhancement, -> enhancement.
l. 98: The utilization of ANN models as fitness functions in optimizations using GA ... ? This is very vague. Surely, you only want to use some output of ANN!
l. 130: deviance -> deviation? (also in Table 1)
(i stopped checking after this. There are too many).

Author Response

Comments of Reviewer 1:

The paper studies the use of Artificial Neural Network (ANN) in combination with Genetic Algorithm (GA) to find optimal combinations of reinforcements to have ultra-high molecular weight polyethylene (UHMWPE) composites with superior mechanical and tribological properties. The studied topic is exciting, and the experiment results seem interesting. However, the whole paper is let down by poor writing and presentation. It can only be reconsidered after a significant revision. The following are some points to consider to improve the paper:

1) The introduction is not clear. The problem is adequately described but current state-of-the-art approaches, what is proposed, and its novelty are all in a blur. The introduction is quite long but has only two paragraphs. One of them consists of 78 lines. The other is also 30 lines. The author might want to break them up into much smaller paragraphs and make the idea clear in each one of them.

Response: As suggested, the introduction section is made clear by breaking them into multiple paragraphs with clarity in each. The novelty in the approach of designing hybrid UHMWPE composites with multiple reinforcements combining micro- and nano-particles is highlighted in the last few lines of the introduction section.

2) The descriptions of methods are also not clear or not adequate. GA is only described by words without a main reference. It is unclear what are its inputs and outputs and how it is connected to the ANN models (what outputs of ANN are used as the input of GA). Nowadays, DNN is much more popular. Even when people use the term ANN they might still use several hidden layers. So please make it clear that you are using only one hidden layer. Also, I believe you have 5 ANN models because you have one model for each output. It was not clear from my first time reading the paper.

 Responses:

• GA is described in detail under the section 2.2.3 with references.
• The detailed list of inputs and outputs used in the present work are given in the table 1 and the same has been discussed briefly in the section 2.1. The connection of inputs and outputs via hidden layer are shown in figure 1 through the ANN configuration. The objective functions generated from the ANN models are used for the optimization work. This has now been explained at the end section 2.2.1 and at the beginning of 2.2.3.
• DNN is normally utilized when the database is big. In the present work, ANN with single hidden layer is chosen due to smaller size of the database. This could efficiently serve the purpose of revealing the correlation of inputs and output. Increase of hidden layers may increase the prediction accuracy of the model with the training data, but the problem of overfitting will definitely increase. Hence it is better to use single hidden layered ANN for small databases.

 

3) The authors never discuss what software/packages are used, what parameters are selected, or the configuration of the machine to run the experiments on.  
Response:

MATLAB^® software was used to perform the ANN modelling and GA based multi objective optimization. The same GA parameters utilized in the earlier study was used for the optimization. It is now added in the 2^nd line of section 4.4.

4) There are many small typos or unusual choices of words. Here are some suggestions for changes:
l.19: the optimum -> an optimal
l.32 : has ->have
l.38: two year latter -> the last two years
l.38: because of -> due to
l. 42: thereof -> therefore?
l. 60: and indeed?
l.69: with -> to
l. 95: enhancement, -> enhancement.
l. 98: The utilization of ANN models as fitness functions in optimizations using GA ... ? This is very vague. Surely, you only want to use some output of ANN!
l. 130: deviance -> deviation? (also in Table 1)
(I stopped checking after this. There are too many).

Response:

All the stated typos and unusual choices of words are rectified and corrected in the entire manuscript.

1. 98: The utilization…: As explained previously, ANN outputs are not used in GA. The ANN models are converted to functions, which are used as the objective functions for the optimization using GA.

Author Response File: Author Response.docx

Reviewer 2 Report

This manuscript studies on the trial to design and development of ultra-high molecular weight polyethylene (UHMWPE) composites by integrating various micro and nanoparticles as reinforcements for enhanced performance of acetabular cups in hip prosthesis. The research direction of this manuscript is meaningful. The problems in this manuscript are as follows.

1. There are problems in the arrangement of this manuscript, which is not strictly implemented according to the requirements of Applied Sciences.

2. Some drawings are not clear, and requirements are revised.

3. The color block data spacing in Figure 6 is too large, resulting in too sparse data display.

4. Why are the contents of lines 319 to 341 bold.

5. The image contents in Figures 7 and 8 are not clear, and the images in the figures are too small.

6. The abstract and conclusion in the manuscript are suggested to be rewritten.

7. The language in the manuscript needs further strict polishing.

Author Response

Comments of Reviewer 2:

This manuscript studies on the trial to design and development of ultra-high molecular weight polyethylene (UHMWPE) composites by integrating various micro and nanoparticles as reinforcements for enhanced performance of acetabular cups in hip prosthesis. The research direction of this manuscript is meaningful. The problems in this manuscript are as follows.

1. There are problems in the arrangement of this manuscript, which is not strictly implemented according to the requirements of Applied Sciences.

2. Some drawings are not clear, and requirements are revised.

Response: The clarity of all the drawings is improved.

 

3. The color block data spacing in Figure 6 is too large, resulting in too sparse data display.

 

Response: The clarity of the images is refined a bit. The size of each plot is increased.

4. Why are the contents of lines 319 to 341 bold.

 

Response: The original submitted manuscript did not have the mentioned lines in bold. This may be occurred while formatting the manuscript from the journal’s end. It is now corrected to normal text.

 

5. The image contents in Figures 7 and 8 are not clear, and the images in the figures are too small.

 

Response: The clarity of the mentioned figures is improved.

 

6. The abstract and conclusion in the manuscript are suggested to be rewritten.

 

Response: As per the suggestion, the abstract and conclusion are refined.

 

7. The language in the manuscript needs further strict polishing.

 

Response: The language in the manuscript is refined.

 

Author Response File: Author Response.docx

Reviewer 3 Report

It is a good numerical analysis work that investigate ANN outcomes. However, there are many abbreviations that is not clearly defined. I would suggest to list all abbreviations at the beginning of this manuscript.

The manuscript need to be written in better way, that makes the reader easy to follow and check by native English speaker.

Author Response

Comments of Reviewer 3:

It is a good numerical analysis work that investigates ANN outcomes. However, there are many abbreviations that is not clearly defined. I would suggest to list all abbreviations at the beginning of this manuscript.

The manuscript need to be written in better way that makes the reader easy to follow and check by native English speaker.

Response: A list of abbreviations is now added. The whole manuscript is refined in a better way to follow the content with an improvement in the language.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

 

I do not see the response at the end of this comment in the revised paper. The authors also need to indicate whether they develop the whole code or use any package/code and cite accordingly.  Parameters also need to be provided to improve reproducibility. 

Response:

MATLAB^® software was used to perform the ANN modelling and GA based multi objective optimization. The same GA parameters utilized in the earlier study was used for the optimization. It is now added in the 2^nd line of section 4.4.

Author Response

The use of Neural Network toolbox of MATLAB® software is mentioned in the last part of the Section 2.2.1. The utilization of GA multi objective optimization toolbox of MATLAB® software is mentioned in the last part of the Section 2.2.3 along with the GA parameters shown in Table 2.

Reviewer 2 Report

This manuscript studies on the trial to design and development of ultra-high molecular weight polyethylene (UHMWPE) composites by integrating various micro and nanoparticles as reinforcements for enhanced performance of acetabular cups in hip prosthesis. The research direction of this manuscript is meaningful. According to the comments of reviewers, the authors made appropriate revisions to the manuscript, making it more perfect.

Author Response

The authors thank the reviewer for his/her positive comments.