Machine Learning Approaches for Predicting Risk of Cardiometabolic Disease among University Students

Round 1

Reviewer 1 Report

The paper introduces a machine-learning model for early cardiometabolic disease detection, comparing five classifiers on 310 patient records. Logistic Regression excels among males (91% accuracy), while Support Vector Machine and Logistic Regression lead for females (87% accuracy). The study highlights a novel "Risk Level" feature derived from fuzzy logic, enhancing interpretability. Recommendations include clarifying the fuzzy logic approach and discussing clinical implications.

Recommendations for Improvement:

• Provide more details on the reasoning behind the choice of the fuzzy logic approach for deriving the "Risk Level" feature. Elaborate on how it enhances interpretability and discriminatory power.
• Consider adding information on other evaluation metrics (e.g., precision, recall, F1 score) to provide a more comprehensive understanding of classifier performance.
• It would be beneficial to include a discussion on the clinical relevance and implications of the study's findings, connecting the machine learning results to potential real-world applications in cardiometabolic disease management.

Addressing the recommended points will enhance the clarity and completeness of the paper, making it more impactful for readers in both the machine learning and medical communities.

Minor edition is required. 

Author Response

1. Provide more details on the reasoning behind the choice of the fuzzy logic approach for deriving the "Risk Level" feature. Elaborate on how it enhances interpretability and discriminatory power.

• Response: Thank you, we appreciate your feedback to improve the work: additional details regarding the utilization of fuzzy logic have already been provided in section ( 4.1.2 Feature Engineering)

2. Consider adding information on other evaluation metrics (e.g., precision, recall, F1 score) to provide a more comprehensive understanding of classifier performance.

• Response: Thank you for your comment: The precision, recall, and F1 metrics are described in sections 5.1.2, 5.1.3, and 5.1.4, respectively, and more information is provided on these evaluation metrics in section 5.2.

3. It would be beneficial to include a discussion on the clinical relevance and implications of the study's findings, connecting the machine learning results to potential real-world applications in cardiometabolic disease management.

• Response.

Thank you for directed us toward this valid point. The discussion part has been highlighting this point as suggested.

4. Addressing the recommended points will enhance the clarity and completeness of the paper, making it more impactful for readers in both the machine learning and medical communities.

Response.

We agree and your valued comments are highly appreciated

Comments on the Quality of English Language

Minor edition is required.

Response

Done thank you 

Reviewer 2 Report

This manuscript devotes half of its content to reviewing literature and the remaining portion to outlining a well-established machine approach. Considering its structure, it might be more suitable as a review paper. Furthermore, crucial aspects of the machine learning approach, such as dataset splitting and distribution of the target class, are not explained. Hence, I cannot recommend this manuscript for publication.

Author Response

This manuscript devotes half of its content to reviewing literature and the remaining portion to outlining a well-established machine approach. Considering its structure, it might be more suitable as a review paper. Furthermore, crucial aspects of the machine learning approach, such as dataset splitting and distribution of the target class, are not explained. Hence, I cannot recommend this manuscript for publication.

Response: In response to the respected reviewer's comments, the updated manuscript has addressed all the concerns raised. The revised version now incorporates all the reviewers' comments and meets their expectations. Specifically, the manuscript has been adjusted to ensure a more balanced distribution of content between the literature review and machine learning approach sections. Additionally, crucial aspects of the machine learning methodology. These modifications have been made to align the manuscript more closely with the expectations for publication. Meanwhile, we disagree with the reviewer regarding considering this paper as a review article since it contains data of participants and methods.

Reviewer 3 Report

Dear authors. I appreciate the opportunity to have been able to review your work. Indeed, early diagnosis in diseases such as cardiometabolic diseases is essential to prevent complications and treat the condition.

I would like to send you a series of recommendations and comments after reviewing your work:

- The objective of the work is not clearly specified. I think that, both in the abstract and at the end of the introduction, they should indicate this more clearly.

-I have had to read the document several times to get a common thread on the topic. It has been really difficult for me to understand your work well, which is why I suggest that you organize the information better so that the reading of the work is simpler and more easily understood.

- I have not found the type of study that is being carried out and it is essential that it appears in the document.

- The introduction, from my point of view, is too long and perhaps requires subsections for better understanding.

- They indicate in the document that they use "pre-existing data", where did they obtain this data? When was it collected? By whom was it collected? It is very important that this information appears clearly.

- I consider that figure 3 does not provide any information and is dispensable.

- They must review the tables. They interchangeably use semicolons to separate numbers. Unifying criteria is important.

I hope that my comments help you so that the manuscript gains quality.

Thank you so much

Kind regards,

Author Response

Comments and Suggestions for Authors

Dear authors. I appreciate the opportunity to have been able to review your work. Indeed, early diagnosis in diseases such as cardiometabolic diseases is essential to prevent complications and treat the condition.

I would like to send you a series of recommendations and comments after reviewing your work:

1- The objective of the work is not clearly specified. I think that, both in the abstract and at the end of the introduction, they should indicate this more clearly.

Response: Thanks for highlighting this point, the objective of the study is now explicitly stated in both the abstract and introduction.

2-I have had to read the document several times to get a common thread on the topic. It has been really difficult for me to understand your work well, which is why I suggest that you organize the information better so that the reading of the work is simpler and more easily understood.

Response: Thank you for your valuable suggestion. We have already restructured the content by dividing the "Introduction" section into two sections: "Introduction" and "Literature Review." Additional information about obesity, overweight, cardiometabolic diseases, and obesity among adolescents in Saudi Arabia has been incorporated into the introduction section, enhancing clarity and providing more comprehensive information.

3- I have not found the type of study that is being carried out and it is essential that it appears in the document.

Response:Thanks for your comment, the type of study is now emphasized by clearly stating the study's aim, which is to construct an artificial intelligence model for predicting the likelihood of developing a cardiometabolic disease.

4- The introduction, from my point of view, is too long and perhaps requires subsections for better understanding.

Response: We appreciate your feedback to improve the work, The content has been reorganized by splitting the "Introduction" section into two parts: "Introduction" and "Literature Review."

5 - They indicate in the document that they use "pre-existing data", where did they obtain this data? When was it collected? By whom was it collected? It is very important that this information appears clearly.

Response: Thank you for your suggestions. Additional details about the collected dataset have been included in Section 3.1.

6- I consider that figure 3 does not provide any information and is dispensable.

Response: Thank you for your comment. Figure 3 illustrates the overall model architecture,

 In this work, we introduce a novel "Risk Level" feature, derived through fuzzy logic applied to the Conicity Index, as a novel feature, previously unused, to enhance the interpretability and discriminatory power for cardiometabolic disease risk assessment. As the Conicity Index scores indicating CMD risk differ between men and women, two separate models will be developed to address each gender individually.

Therefore, Figure 3 shows the development of two distinct models—one for male and another for female students.

7- They must review the tables. They interchangeably use semicolons to separate numbers. Unifying criteria is important.

Response: Thank you for your feedback; the tables have been revised accordingly.

Reviewer 4 Report

General comments

=============

As a reviewer, I commend the effort and structure of your manuscript titled “Machine Learning Approaches for Cardiometabolic Disease Risk Prediction among University Students.” The clear and logical organization is notable, yet there are several areas that need attention to elevate the quality and impact of your work. I have outlined specific areas of improvement below:

Specific comments

=============

Major comments

---------------------

1. Title and Abstract Consistency* There is a notable discrepancy between the target population mentioned in the title (university students) and the abstract (patients dataset). This inconsistency needs to be resolved for clarity and accuracy.

2. Abstract - Definition and Data Details: The abstract should include a clear definition of cardiometabolic disease to set a precise context for the study. Furthermore, it is crucial to specify the number of records analyzed, distinguishing between those with and without the outcome of cardiometabolic disease.

3. Introduction - Contextual Clarity: The introduction should explicitly delineate what is known and unknown in the field, and why this study is necessary. The referenced articles need to be integrated more effectively, not just summarized. Particularly, a focus on adolescents and cardiometabolic disease in relation to your study’s target and outcomes is essential.

4. Introduction - Obesity Context: Expand the introduction to include information about obesity among adolescents in Saudi Arabia and on an international scale, to provide a broader context for your study.

5. Methodology - Data Cleaning Details: The methods section should elaborate on the data cleaning processes used, specifically how missing data were handled and the rationale behind these choices.

6. Methodology - Definition and Data Distribution: As in the abstract, include a clear definition of cardiometabolic disease in the methodology. Also, detail the distribution of records with and without the cardiometabolic disease outcome, which is crucial for understanding the study's scope.

7. Methodology - Gender-specific Models: Explain the scientific reasoning behind the decision to create different models for males and females. This justification is essential for understanding the methodological choices and their implications.

8. Discussion - Direct Result Analysis and Contextualization: There was no discussion part, Therefore, the discussion should directly address the results of the study. It should not only interpret these results but also compare them with previous studies, suggest future research directions, and acknowledge any limitations of your study.

Addressing these points will significantly enhance the manuscript in terms of clarity, coherence, and scientific contribution. Your study has the potential to make a meaningful impact in the field, and these improvements will help in achieving that.

Please refer the previous quality of English rating.

Author Response

1. Title and Abstract Consistency* There is a notable discrepancy between the target population mentioned in the title (university students) and the abstract (patients dataset). This inconsistency needs to be resolved for clarity and accuracy.

Response; We appreciate your feedback on improving the work. The inconsistency has been addressed, and now both the title and abstract are harmonized, accurately reflecting the target population as university students. 

2. Abstract - Definition and Data Details: The abstract should include a clear definition of cardiometabolic disease to set a precise context for the study. Furthermore, it is crucial to specify the number of records analyzed, distinguishing between those with and without the outcome of cardiometabolic disease.

Response: Thank you for your valuable input. The abstract has been revised in accordance with the provided feedback.

3. Introduction - Contextual Clarity: The introduction should explicitly delineate what is known and unknown in the field, and why this study is necessary. The referenced articles need to be integrated more effectively, not just summarized. Particularly, a focus on adolescents and cardiometabolic disease in relation to your study’s target and outcomes is essential.

Response: Thank you for your valuable suggestion. We have reorganized the content by splitting the "Introduction" section into two parts: "Introduction" and "Literature Review." Additional details regarding obesity, overweight, cardiometabolic diseases, and obesity among adolescents in Saudi Arabia have been integrated into the introduction section, improving clarity, and offering more comprehensive information. The literature review section has been enhanced, and a paragraph has been included after the table to highlight the focus of this study in comparison to existing research.

4. Introduction - Obesity Context: Expand the introduction to include information about obesity among adolescents in Saudi Arabia and on an international scale, to provide a broader context for your study.

 Response: Thank you for your comment. Additional information about obesity among adolescents in Saudi Arabia and on an international scale has been integrated into the introduction section, improving clarity and offering a more comprehensive overview.

5. Methodology - Data Cleaning Details: The methods section should elaborate on the data cleaning processes used, specifically how missing data were handled and the rationale behind these choices.

 Response: Thank you for this comment, more details have been incorporated into section (4.1.1 Data Cleaning)

6. Methodology - Definition and Data Distribution: As in the abstract, include a clear definition of cardiometabolic disease in the methodology. Also, detail the distribution of records with and without the cardiometabolic disease outcome, which is crucial for understanding the study's scope.

Response: Thank you for your feedback, cardiometabolic diseases are introduced at the beginning of the methodology section ( i.e. section 4. Methodology).

The distribution of the records is highlighted in subsection (4.1.3 Categorical Encoding) and also in Figures 5 and 6.

7. Methodology - Gender-specific Models: Explain the scientific reasoning behind the decision to create different models for males and females. This justification is essential for understanding the methodological choices and their implications.

 Response: Thank you for pointing this out: In our work, Conicity Index (C index) is used to predict the risk of cardiometabolic disease CMD. Since the Conicity Index scores indicating CMD risk vary between men and women, two distinct models were constructed to address each gender separately.

This justification has been incorporated into section (4.1.2. Feature Engineering)

8. Discussion - Direct Result Analysis and Contextualization: There was no discussion part, Therefore, the discussion should directly address the results of the study. It should not only interpret these results but also compare them with previous studies, suggest future research directions, and acknowledge any limitations of your study.

Response: Thank you for your suggestion. Section 5 has been renamed to "Experimental Results and Discussion," and the section's content has been updated based on the provided comments

Round 2

Reviewer 2 Report

The authors have provided reasonable responses to my comments, and have revised the paper accordingly. I'm now happy to recommend acceptance.

Reviewer 3 Report

Dear authors. I have carefully read your comments and improvements to the manuscript. I consider that it has been enriched by the reviewers' comments. I congratulate you for your work.

Best regards

Reviewer 4 Report

As a reviewer, I commend the effort and structure of your manuscript titled “Machine Learning Approaches for Cardiometabolic Disease Risk Prediction among University Students.” The clear and logical organization is notable. Almost all responses were reasonable.