Tabular Data Generation to Improve Classification of Liver Disease Diagnosis

Round 1

Reviewer 1 Report

Title: Tabular Data Generation to Improve Classification of Liver 2 Disease Diagnosis

In this work, the authors aim to assess the performance of various machine learning algorithms to decrease the cost of predictive diagnoses of chronic liver disease. They employed five logistic regressions, the K-Nearest neighbor, the Decision Tree, the Support Vector Machine, and ANN algorithms. Also, they examined the effects of the increased prediction accuracy of GANs and SMOTE.

However, the study is fundamentally problematic. The authors are definitely failing in the LR section. I highly recommend that authors read the LR sections of quality papers. It is also unclear how the paper will act as a bridge in the literature. Also, an important issue is that the references section contains only historical works. It is impossible to understand why the most recent studies are not included in this section. In general, the paper fails to be accepted.

Author Response

We would like to thank reviewer for sharing their valuable time in reviewing our manuscript. We would like to express our sincere thanks to the reviewer for the constructive and positive comments. Please find point-to-point clarification/reply for the concerns raised. All modifications made in the manuscript are highlighted in yellow color.

Author Response File: Author Response.pdf

Reviewer 2 Report

In the introduction, section 1 and line no 100, the author mentioned, "In this paper, six classification algorithms are applied and evaluated with and without augmentation on ILPD (Indian Liver Patient Dataset) dataset, " but the dataset contains only  583 entries. How does the author use GAN ? As it is very less no of records than what is use to apply the machine learning algorithms like (DT, SVM,K-NN .etc ) . The real problem with insufficient data lies in the fact that with less data, variance increases. Variance, which can easily be defined as the variability of model prediction for a given data point or a value that tells us how the data is spread. High variability in models means that the model will fit the training data perfectly, not for testing data.

So the author needs to give proper clarification about it.

In section 7 and line no.384, the author discussed the common 10-fold cross-validation approach used, but they have not discussed it thoroughly at each fold what is the training accuracy as well as the testing accuracy?

In the conclusion section, the author claims the novelty but makes no such comparison of the current approach to the approach which is available in the literature. 

Neither the SOTA approach was used in the literature nor even in the discussion section.

No comparison done after SMOTE and before SMOTE?

Rather than focusing more on the dataset, sampling approach, and robust machine learning algorithms, the author has focused more on Exploratory Data Analysis.

It seems haphazardly collected and put in a common context.

 

 

 

Author Response

Thanks for sharing your valuable time in reviewing our manuscript. We would like to express our sincere thanks to the reviewer for the constructive and positive comments. Please find point-to-point clarification/reply for the concerns raised. All modifications made in the manuscript are highlighted in yellow color.

kindly see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

This version of the paper can be accepted after minor revision. Some your methods-related papers must be discussed in the paper, including Training multilayer perceptron with genetic algorithms and particle swarm optimization for modeling stock price index prediction; A comparative analysis of K-Nearest Neighbour, Genetic, Support Vector Machine, Decision Tree, and Long Short Term Memory algorithms in machine learning; Landslide susceptibility mapping based on the germinal center optimization algorithm and support vector classification; A novel approach for feature selection and classification of diabetes mellitus: Machine learning methods.

 

 

Author Response

Thank you for taking the time to review our paper. We appreciate your feedback and are glad to hear that the paper can be accepted after minor revisions. We will make sure to include the suggested methods-related papers in the discussion section of the paper. Please refer to the highlighted text in yellow in revised manuscript.

Reviewer 2 Report

The authors addressed the questions asked in the first round, but still, I am not convinced by Table 7. comparison with other research work. The papers which were referred to are either conference papers or not good journals. Many articles are there on the internet of sources that the author has not selected. 

Authors are suggested to select a good journal and compare it .

In the last section, the reference is terrible style. 

 

Author Response

Thank you for taking the time to review our paper and for providing your constructive feedback. We appreciate your comments and suggestions, and we will strive to improve the quality of our work based on your recommendations.

Regarding the Table 7 comparison with other research work, we understand your concerns and we will revise the table to ensure that it includes articles from reputable journals and provides a comprehensive comparison with relevant studies.

We apologize for the reference style in the last section, and we will make sure to follow the proper citation and referencing guidelines in the revised version. Please refer to the highlighted text in yellow in revised manuscript.

Author Response File: Author Response.docx