A Comprehensive Study on Healthcare Datasets Using AI Techniques

Round 1

Reviewer 1 Report

Please find the attached file.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This study aims to analyze and improve the quality of healthcare datasets by using NARX and supervised machine learning methods. The models employed are evaluated with the means of characteristic plots and the sensitivity data, the Regression procedure was used to obtain specificity and positive predicted values.

This research work is great however I have some suggestions:

The authors of this study refer:“Data Availability Statement: The article contains the original contributions made for this study; 458 further questions should be addressed to the corresponding author”. However, the reader or researcher should at least consult the supplementary file containing the original dataset and then all processed data (the data used to support the findings of this study). As the volume of the original dataset obviously is very large, alternatively, the exact title and URL link of the dataset as it comes from healthdata.gov can at least be given.

It would also be good to mention the software used to process the data and display the graphs.

It would also be useful to cite other related work and make a model comparison between the rest and the method used here.

Finally, some typographical and wording errors (e.g. lines 123, 126, 128, 143, 194, 238, 444, 452) should be corrected.

Good luck

 

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors,

   Thank you for your manuscript submission to MDPI Journal of Electronics. This is a survey paper aimed at comprehensive investigation of AI-oriented approach applied to healthcare datasets. While it studied NARX, some machine learning methods and provided some performance analysis, the current version contains some problematic issues, which are quite explicit. I suggest the authors provide one round of major revision, the major and minor problems suggested to be fixed, are listed as follows:

   1) The Abstract session lacks keynote quantitative scores on the concluding remarks, and it looks a bit generic. Please limit its lengths to 150~200 words.

   2) Approach: a good survey paper may contains 100+ references, however, the authors only cited 29. The machine learning schemes are not restricted to supervised ones, semi-supervised, weakly supervised approaches as well as a few unsupervised methods could be within your scope of investigation. Also, I don't think this paper addressed latest deep learning and reinforced learning methods for AI-techniques in healthcare. Please update. 

   3) Related problems in the content:

       a) Introduction: it lacks a short session to discuss your contributions of your work (typically in 3-4 mainfolds in summary). Also, I don't think the first 3 paragraphs qualify a good review for prior study. Please conside re-writing.

       b) Section 2 should be "Related work", and the methodology should not just be confined to supervised machine learning, and the current subsection also is quite thin, lacking specific details on mathematical modeling and its variations. The bottleneck of current study, importance of your review can be further addressed, in my opinion.

       c) Comparative study: the subsection 3.1 can be expanded, and the data source as stated in subsection 3.2, can be shifted to Section 4, with a better title of "Experiments and Results".

       d) Figures and Tables: Most of them looks fine. The positioning of Fig. 4 lokos awkward, which should be shifted and put in the right location (middle of the page, no surrounding of paragraphs on statements). The required interval before and after each diagram should be shaped according to the MDPI template reuqirement.

       e) Discussion and Conclusions look just OK, while the last paragraph on future work can be further expanded with a little bit details on the summary of research challenges and possible solutions with respect to your study.

   4) References: the citations looks a bit weak, which should be further addressed. The cited journals should be shaped in abbreviated formats. A lot more state-of-the-arts in Years 2020-2022 need to be supplemented in the upgraded version. Besides, those irrelevant notations such as ''  ' ' or vol., no.  ", should be deleted (please check published articles in MDPI journals).

   5) Minor problems: the use of English should be significantly improved. The current version has quite a lot of parts (which are hardly readable), please invite a native English speaker to help. The hyphenating issues between adjacent lines should be fixed. Meanwhile, some minor shadows inside Fig. 1 and other subsequent graphs, are supposed to be eliminated. 

   Once again, thank you, we wish you the best of luck on paper edits towards next round of double-decision process. Take care!

Stay well,

Best regards,

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

The purpose of the manuscript is to propose a method for the different characteristics that allocate to or contact all sources while using artificial intelligence techniques to consider the non-unique features and calculate the similarity indices. Authors also evaluate using characteristics plots and sensitivity. In my opinion, the manuscript is suitable for publication in the journal “Electronics” only after solving the mentioned minor issues.

Minor points:

1.      Authors are suggested to place summarize the result in one or two sentences and add them at the end of the abstract.

2.      Is figure 1 the author’s creation or taken from another place?

3.      Authors need to do a better literature survey, very few works were mentioned from 2021 and 2022.

4.      In lines 117 -120 author mentioned “algorithms”, authors need to provide prominent algorithm names with citations here.

5.      What is the rationale behind choosing NARX neural network as an example here? Authors need to explain that before explaining NARX neural network.

6.      Equations 1 and 2 are missing citations. Unless the authors propose that equation, they need to be cited.

7.       Does changing the training portion (60%) in line 139 to let's say 80% make any significant change in performance (figure 2)?

8.      Figure 4 labels are very small.

 

9.      The discussion should summarize the outcome of the numerical study section and the authors should elaborate on the significance of the analysis of existing studies. Authors should work on the Discussion section and elaborate it significantly. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors,

The manuscript entitled: A Comprehensive Study on Healthcare Datasets Using AI-Technique

As a second review, I take over that the revised manuscript has improved significantly both in style and substance. I also reviewed the “Response to Comments” file provided by the authors and it seems that the authors successfully responded to all the reviewers’ suggestions and other comments. The methodology parts are now much clear. Most part the concerns and comments were addressed in appropriate ways.

However, the conclusion part has not been modified. The authors should give details of academic and practical implications, a discussion of results, and future research for practitioners and academicians in the conclusion section.

I assumed that the paper is now publishable after revise of the conclusion part and submitting the final version. Please review a grammatical error in the manuscript.

 

I appreciate the authors’ time and efforts in the revision of the research.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors, 

   Thanks a lot for your carefully edited second version for review. I think the quality of this paper has been significantly improved with respect to my prior comments, hence, I have recommended your paper as acceptance with minor edits. The remaining problematic issues must be addressed are listed as follows:

   a) The introduction part is very short, and the authors asserted that "no study" is related to complex healthcare databases, which is quite arbitary. I think the statements at Lines 52-57 should be replaced with a main summary of your main contributions (with a little bit specific details).

   b) Figures and Tables: The visual quality of all the figures should be further enhanced, it is better to attach the original version (with highest resolution). Besides, stop crossing over one table in two adjacent pages (which are not professional); the title of Tables looks too simple to follow, please be more specific. These updates should be applied in the accepted version.

   c) Discussion section looks fine, while a few minor defects at Line 454 and Line 469 should be fixed. State with "Fig. 5(a)" and "Fig. 6(a)" instead of the current shape. If any results from ablation study in support of your main conclusion, they are suggested for supplementation. Thanks a lot!

   d) Conclusion: The third paragraph specified some future work, while I think the professional one should be comprised of at least 3 parts: main summary of their work, opening problems to be solved (or a summary of research challenges), and brief summary of research orientations on future study. Thanks a lot!

   e) References: A few aspects of revisions still have to be addressed: a) apply abbreviated style on the title of some journals;  b) Some latest publications in Years 2020-2022 which are similar / parallel to your study area (some deep learning based AI schemes on healthcare datasets and clinical data analysis), are recommended to be supplemented; c) Citations on each of the highly visible conference proceedings should be loaded with time duration, location, and page numbers. Please follow the template and update with required edits. Thanks very much!

   f) This version also contains some minor issues, which are suggested for editing: 

    1) The literal quality is acceptable, while use of English should be improved in the updated version. A few minor grammatical issue need to be fixed. Please conduct proofreading on the related context carefully. For instance, what does "Were implemented" mean at Line 381? 

    2) Align the size of figures (a few of them are a bit too large) and the title of some figures must be more specific. Align the intervals on line of tables. Check the MDPI template for the required font size.

    3) Fix the hyphenating issues in the whole context. When you are using MS word or Latex, please avoid hyphenating a word (which currently appears multiple time at the end of some lines to cross-over two adjacent lines). The MDPI online template has the options to adjust that. 

    4) Remove some redundant indenting issues at the beginning of some paragraphs, and delete some redundant half-spacing in the context. 

   In sum, I wish you the best of luck for updated manuscript coming into acceptance. Again, thanks for your manuscript submission and updating.

Stay well,

Yours faithfully,

Author Response

Please see the attachment

Author Response File: Author Response.pdf