Cervical Cancer Diagnostics Using Machine Learning Algorithms and Class Balancing Techniques

Round 1

Reviewer 1 Report

I read the paper and found it interesting and outstanding 

I suggest to do rewriting introduction and discussion sections considering more literature review 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Cervical Cancer Diagnostics Using Machine Learning Algorithms and
Class Balancing Techniques by Matko Glucina et al.

The authors use publicly available cervical cancer data with 36 inputs
and 4 outputs. Various Artificial Intelligence and Machine Learning
approaches are considered with the purpose of early detection of this
for of cancer. Class balancing is employed. The approaches are
compared and discussed.

The hypotheses set forth are meaningful. Although I am not well-versed
with the literature, the study is probably novel enough to warrant
publication. A positive aspect of the paper is that a variety of
approaches are tested and compared.

The most important deficiency is that while lots of numerical data and
charts are generated, the discussion is too brief. More discussion of
the results would be better.

I could not find substantial fault with the paper. However, given the
paper is only marginally within my area of expertise I am not highly
confident in my review.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

I am happy to review this paper. 

1. As you mentioned that the data resource is from "publicly available cervical cancer data collected on 859 female patients" on page 1, line 9 as well as on page 6, line 195, I have checked the data set accordingly and found that this dataset focuses on the prediction of indicators/diagnoses of cervical cancer. The features cover demographic information, habits, and historical medical records. 

The dataset was collected at 'Hospital Universitario de Caracas' in Caracas, Venezuela. The dataset comprises demographic information, habits, and historical medical records of 858 patients. I am feeling that It is not a very clean dataset. I am afraid that I can't trust it.

2. Could you give the definition for what is AUC and MCC in your abstract on page 1 line 20, as you use them on page 12, Table 5? 

Is it means that for "Area Under The Curve (AUC)" and "Matthew's correlation coefficient (MCC)" respectively?

3. Similar question on page 3, Table 1, the abbreviations of "true positive rate (TPR)" and "true negative rate (TNR)" respectively for the confusion matrix are not definite in your paper.

All these issues make readers don't want to read your paper completely patiently.

4. Could you have some clinical collaborations to get some real data for your training and testing in your AI/ML? It will be much greater convenience to the readers who really want to know if you do so.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

This manuscript deals with the use of AI-based algorithms in the early diagnosis of cervical cancer. The introduction section is too long. The authors extensively describe cervical cancer diagnosis. However, there is no indication about the success rate of cervical cancer screening. There is no information on how many patients with cervical cancer do not attend the screening?  What percentage of cervical cancers can be prevented? What is the sensitivity and specificity of PAP, LBC, HPV? These are in combination with colposcopy highly reliable methods in preventing cervical cancer and the authors do not justify the use of AI-based algorithms in clinical practice.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Thank you for addressing my comments.

The last part in your Conclusions. you use the same "Furthermore" twice in the same area. Could you find another word to replace one of them? 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

I have no further comments

Author Response

Please see the attachment.

Author Response File: Author Response.pdf