Feature Selection of Microarray Data Using Simulated Kalman Filter with Mutation

Round 1

Reviewer 1 Report

Gist/Summary:  The work proposes Simulated Kalman Filter with Mutation (SKF-MUT) for microarray data's feature selection  which the authors claim that it was NOT employed earlier by any. The algorithm enhances  classification accuracy of ANN and is based on a metaheuristics optimization algorithm wherein the authors used benchmarked  datasets  comprising diffuse large b-cell lymphomas. They use both binary and diffused datasets and employ Accuracy using confusion matrix 

 

I couldn't find the supplementary datasets and data availability to check the diffused/binary datasets

 

Figure 6 must be redrawn as it is not clear 

Perhaps, it is the first approach is an udner-statement as there are methods that have been applied: https://pubmed.ncbi.nlm.nih.gov/17065158/

 

Minor but essential

Made subtle grammatical corrections

The methods could have a pictorial methodology clubbing Figure 1 and 3

 

 

Scores on a scale of 0-5 with 5 being the best 

 

Language: 4

Novelty: 3.5

Brevity: 4

Scope and relevance: 3

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

In this research, authors aim to enhance the classification accuracy of artificial neural network by applying Simulated Kalman Filter with Mutation for microarray data’s feature selection. In my opinion, the subject of study is important. But the paper in some aspects is unclear. Minor revisions are required before publication.

1.     Abstract, line 18: Benchmark datasets were used…brain tumor, nine tumors and 11 tumors. This is indeed an involved sentence. What does nine tumors and 11 tumors mean?

2.     The accuracy of the method should be clearly reported in the abstract and the related section of the manuscript.

3.     Have the authors considered the ethical issues? Does the data collection process in compliance with relevant requirements?  

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper proposed Simulated Kalman Filter with Mutation (SKF-MUT) for microarray data's feature selection to enhance the classification accuracy of ANN. The algorithm is based on a metaheuristic’s optimization algorithm inspired by the famous Kalman filter estimator. The mutation operator is proposed to enhance the performance of the original SKF in the selection of microarray features. Benchmark datasets were used, which comprise of diffuse large b-cell lymphomas (DLBCL), prostate cancer, lung cancer, leukemia cancer, “small, round blue cell tumors” (SRBCT), brain tumor, nine tumors and 11 tumors. The accuracy is taken as the performance measurement by considering the confusion matrix. Based on the results, SKF-MUT effectively selected the number of features needed leading towards higher classification accuracy.

 

1)    The overall writing has some formatting issues, like wording and spacing. I suggest the authors check the grammar and avoid any typos. More importantly, the writing needs improvement.

2)    The method part is lack of details. More detailed descriptions are needed to explain this part.

3)    The results are not quite sufficient. More discussions on the result part are needed. Moreover, I would suggest the authors discuss using different approaches (e.g., PMID: 36545790) as future perspectives, to further investigate the conclusions.

N/A.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

The article is devoted to the problem of feature selection of microarray data. Authors propose to use simulated Kalman filter with mutation to improve the classification accuracy of neural networks. Cancer research is a relevant theme, and the article has strong practical importance. However, the manuscript has gaps in scientific novelty.

Introduction section lacks information about DNA microarrays. I suppose, additional information in the Introduction section will be useful.

The article contains errors and typos such as “…to select features of two different type…”, “…the incorporation of feature section is observed to be able to improved classification accuracy…”, “parameter-less”, “This parameter setting is set…” and so on.

The choice of accuracy as a measure of classification quality is not justified. The dataset is not described enough. Was there a class imbalance? In this case, it is unacceptable to use accuracy as a measure of classification.

In Tables 2-4 the best solution presented. However, the best solution is not indicative, it would be better to use the mean or median combined with variation. Statistical analysis is declared in the text, but its results are not presented.

Some other comments

Line 20 - In the Abstract, what meant by “nine tumors and 11 tumors”?

Line 49 – Abbreviation VLSI (and following ones) should be translated.

Line 77 – Different writing “section” and “Section”.

Line 78 – The dot is absent.

Line 162 - ??(t) – i is not specified.

Line 175 – In (2) Z?(t) is not specified. Formulas must be punctuated throughout the text.

Line 181 – Equations (2) and (3) are the same.

Line 188 – Figure 1 and Figure 4 quality is poor.

Line 193 – Formulas (7) and (8) are not denoted. What is d? What is randn?

Line 325 – Quality of Figure 6 is unacceptable.

Line 357 - What is “z” in Table 4?

References should be formatted according to journal requirements.

The quality of the English language needs improvement

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I am satisfied with the changes and responses rendered!

Reviewer 3 Report

The authors have addressed my concerns. This manuscript is ready to be accepted.