Cancer Detection Using a New Hybrid Method Based on Pattern Recognition in MicroRNAs Combining Particle Swarm Optimization Algorithm and Artificial Neural Network
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis topic is interesting and seems to be applicable. However, I think there are some concerns to be resolved carefully at this stage. Please find my comments in the following:
- The paper lacks the following details that are extremely important to make a proper assessment of the contributions that have been done:
- The comparison with other systems is very incomplete. What do you mean by optimization and adaptation between PSO and ANNs ? Which are the characteristics of it? How has it been trained? Which are all the parameters that you are taking into account?
- The Description of Datasets/ the configuration/setting of parameters/hyper-parameters that have been used for the evaluation.
The lack of real-world scenarios, models and evaluations with recent schemes that present similar elements is one of the weak points of this paper.
The computational time should be reported.
The usage of certain abbreviations are inconsistent.
There is no link of code in the paper. Provide github or any other relevant link so that proposed work may be tested.
The details of the conducted experiments can enable the reader to find the critical information for reproducing the same results.
Finally, the paper is not written well, which contains the problem of normativity and many grammatical mistakes and typos.
Comments on the Quality of English LanguageFinally, the paper is not written well, which contains the problem of normativity and many grammatical mistakes and typos.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe study refers to the problem of cancer detection via microRNA analysis. Authors propose a method based on combining Particle Swarm Optimization algorithm and Artificial Neural Network. The area of research is undoubtedly relevant. The study is well written.
The main comments are as follows.
1) The main contributions of research should reflect the scientific novelty that this work brings and represent a solution to those research gaps that have not yet been addressed. Most of the contributions presented in this work have previously proposed solutions. Please, specify the contributions to highlight your own findings.
2) In fact, the literature review of existing approaches contains only three references directly related to the research topic ([36]-[38]). Moreover, these links refer to a period of 7 years ago. The review needs to be expanded to include more recent sources, since the field of machine learning (including image processing in medicine) is developing extremely dynamically.
In addition, existing methods of feature selection in miRNA datasets were not considered.
3) Research design is questionable. The number of samples is extremely low for training the network. Results are not well described. What was the train/test ratio? Are the Tables 1-3 demonstrate results for the test set? Judging by the number of samples, division was not carried out, but the results should be presented for a test set that did not participate in the training of the neural network.
Why accuracy was chosen as a metric? Accuracy is not an informative metric in the case of class imbalance and its use in research is questionable.
Most of the methods used for comparison were developed 7 to 12 years ago. The main results in the analysis of medical images were obtained later. It is necessary to compare the proposed method with more recent state-of-the-art methods.
Some other comments:
- There are many phrase repetitions in the article, like:
“The input layer is the first layer in a network, …, with the input layer being the first.” “ … a model dubbed Xception [ 33 ] … and is referred to as the Xception model”. “They omitted all tumor forms that are sex-specific from consideration … including BRCA, CESC, OV, PRAD, TGCT, UCEC, and UCS, from consideration.” “The fitness function is presented by the following equation, which is the proposed formula of this paper, which is Fitness”.
- Line 115 - Abbreviation “CNN” is not described.
- Authors use the same abbreviation KNN for kernel neural network and k-nearest neighbors. This may lead to misunderstandings.
- Line 206 – ” 2d, where d denotes the size F), which is not computationally optimum” – A part of the sentence is absent.
- Line 226 – “Assessment of the suggested approach This paper will use three GEO…” - A part of the sentence is absent.
- In (1), (2) and (12) X, V and Y are not described.
- Line 274 – “The symbol ni represents..” – i should be subscripted here
- Line 285 – “…means a small batch dispersion and a large batch dispersion” – It is something confusing here
- The quality of Figure 1 is too low.
- Figure 2 should be mentioned in the text before its first appearance.
- Comparison with other methods should be placed in the Results section.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript presents a hybrid model that combines the PSO algorithm and ANN for miRNA pattern recognition. The manuscript is not very well written overall, and I suggest major revision for consideration of publication:
1. The manuscript lacks significant results. There are only two figures in total. One for the architecture, and the other one for validation accuracy. The readers have no idea what the data looks like and what the prediction result looks like. Without these plots, intermediate results and comprehensive analysis, the proposed method’s performance is not justified.
2. It’s hard to believe that the proposed method has much better performance than the cited sources. First of all, as mentioned, there aren’t enough results that demonstrate it. Secondly, what are these cited methods that you are comparing with? Table 4 shows the references of other methods you are comparing with, but what are these methods exactly? What’s the difference? And why is your method better?
3. The manuscript needs extensive English editing. Please see details in the comments about English language.
4. Figure 1 has a very low resolution.
Comments on the Quality of English LanguageOverall, the manuscript needs an extensive English editing, as many sentences are difficult to comprehend. Here are a few examples of language issues that happen in one paragraph (note: it’s not a comprehensive list):
1. Line 194: “By and large, a good..” I don’t understand this sentence, maybe due to grammatical errors.
2. Line 195 - 196: “ well-chosen miRNA may offer…, but unselected.. only expand”. I don’t understand the contradiction here.
3. Line 199: What does “accomplish.. performance” mean?
4. Line 206: “2D, where .. computationally optimum”. This is not a complete sentence.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsAll my raised remarks have been taken into consideration appropriately, thank you.
Comments on the Quality of English LanguageAll my raised remarks have been taken into consideration appropriately, thank you.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsIt is necessary to add a description of F1 Score to Section 4.5, and also add the results of its calculation to Table 7.
Table 6 duplicates Table 5
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript has improved in terms of English, and the authors have added descriptions about some details, but the authors didn’t address my comments 1 and 2 well.
There are still only two figures in the manuscript, without any plots about the data, intermediate results, prediction results, and comparison of prediction results with other methods. This is not a complete research without necessary plots.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 3
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript has improved based on comments.
