Exploring the diagnostic performance of machine learning in prediction of metabolic phenotypes focusing on thyroid function

Hyeong Jun Ahn; Kyle Ishikawa; Min-Hee Kim

doi:10.1371/journal.pone.0304785

Peer Review History

Original SubmissionOctober 24, 2023
13 Feb 2024 Decision Letter - Vijayalakshmi Kakulapati, Editor PONE-D-23-34518Exploring the diagnostic performance of machine learning in prediction of metabolic phenotypes focusing on thyroid functionPLOS ONE Dear Dr. Ahn, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Mar 29 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Vijayalakshmi Kakulapati, Ph.D Academic Editor PLOS ONE Journal Requirements: 1. When submitting your revision, we need you to address these additional requirements. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf. 2. Note from Emily Chenette, Editor in Chief of PLOS ONE, and Iain Hrynaszkiewicz, Director of Open Research Solutions at PLOS: Did you know that depositing data in a repository is associated with up to a 25% citation advantage (https://doi.org/10.1371/journal.pone.0230416)? If you’ve not already done so, consider depositing your raw data in a repository to ensure your work is read, appreciated and cited by the largest possible audience. You’ll also earn an Accessible Data icon on your published paper if you deposit your data in any participating repository (https://plos.org/open-science/open-data/#accessible-data). 3. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match. When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section. 4. Thank you for stating the following in the Acknowledgments Section of your manuscript: [ Hyeong Jun Ahn and Kyle Ishikawa are partially supported by the National Institute of Health (2U54MD007601-36 and U54GM138062). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.] We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form. Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows: [The author(s) received no specific funding for this work.] Please include your amended statements within your cover letter; we will change the online submission form on your behalf. 5. Please include a copy of Table 3 which you refer to in your text on page 18. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: Yes Reviewer #2: Partly Reviewer #3: Yes ******** 2. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Yes ****** 3. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Yes ****** 4. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: Yes Reviewer #2: Yes Reviewer #3: Yes ****** 5. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: 1- The name and information about the database used were not mentioned in the abstract 2- In the introduction, please add the researcher’s contributions. In addition to adding a paragraph at the end of the introduction explaining the structure of the research 3- The working algorithm is not clear at all. In addition, the reasons for choosing them or the difference between them were not explained. Also, please pay attention to the numbering of titles 4-The results for the metrics are presented in a table rather than in text form 5- The discussion is devoid of any mention of the results with their numerical values. It is also preferable to add the conclusions in a separate section 6-Why is deep learning not mentioned as future work? Is it unlikely for certain reasons? 7-References need to be more standardized and organized Reviewer #2: 1. It only used data from a single study/dataset (NHANES), limiting generalizability. Using multiple datasets could strengthen findings. 2. Performance comparisons were limited to 5 commonly used algorithms. Testing more recent deep learning methods may yield better results. 3. Validating models on held-out test data from a later NHANES wave would better reflect real-world use, versus internal validation. 4. Limited demographic/clinical covariates were considered; including more exhaustive potential predictors could impact results. 5. Interactions and non-linear effects between variables were not modeled, which may be important for complex phenotypes. 6. Describe dataset features in more details and its total size and size of (train/test) as a table. 7. Pseudocode / Flowchart and algorithm steps need to be inserted. 8. Time spent need to be measured in the experimental results. 9. Limitation Section need to be inserted. 10. The architecture of the proposed model must be provided 11. Address the accuracy/improvement percentages in the abstract and in the conclusion sections, as well as the significance of these results. 12. The authors need to make a clear proofread to avoid grammatical mistakes and typo errors. 13. Add future work in last section (conclusion) (if any) 14. To improve the Related Work and Introduction sections authors are recommended to review this highly related research work paper: a) Optimizing epileptic seizure recognition performance with feature scaling and dropout layers b) Optimizing classification of diseases through language model analysis of symptoms c) Predicting female pelvic tilt and lumbar angle using machine learning in case of urinary incontinence and sexual dysfunction d) Utilizing convolutional neural networks to classify monkeypox skin lesions e) Hepatitis C Virus prediction based on machine learning framework: a real-world case study in Egypt Reviewer #3: Thank you for the hard work presented. It is nicely written and covers an important topic. Authors could take the advantages of ML algorithms in modeling one of the medical diagnosis applications. However, could provide them with some few comments: 1- Generally, along the sections, some paragraphs with the same information have been duplicated. Try to remove them as much as possible. 2- Authors could provide a section describing the dataset, its complexity (if) and other details that have direct impact on the outcome of the ML models. 3- The paragraph from the line 150 - 166, perhaps this paragraph be moved earlier when authors talk about preprocessing and Table 1. 4- in line 152, there is Table X, is it the Table 1 or another table not being included in the document? 5- Why did not the authors provide a Figure shows the AUROC? 6- Authors could compared their results with some state-of-art works? 7- lines 258 - 260: this is a general advantage of ML that already being presented in the Introduction. Have you developed a real-time ML model? 8- If you believe that the dataset you are using is complex, why you used a regular Neural networks? why you did not implemented the deep learning? 9- How can we reflect the work of MacNell on yours? How can you use their conclusions to support your results? 10- You have mentioned about developing a confusion matrix. Can you show a Figure visualizing them? ****** 6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No Reviewer #2: Yes: Tarek Abd El-Hafeez Reviewer #3: No ******** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. https://doi.org/10.1371/journal.pone.0304785.r001
Revision 1
19 Apr 2024 Author Response Point by point response to reviewer’s comments Note: We have provided our response in blue. The line and page numbers referenced correspond to the tracking version of the manuscript. Reviewer #1: 1- The name and information about the database used were not mentioned in the abstract Thank you for your comment. We utilized the National Health and Nutrition Examination Survey (NHANES) from 2007 to 2012 and made a correction in Abstract. “In this study, we employed various machine learning models to predict metabolic phenotypes, focusing on thyroid function, using a dataset from the National Health and Nutrition Examination Survey (NHANES) from 2007 to 2012. ” 2- In the introduction, please add the researcher’s contributions. In addition to adding a paragraph at the end of the introduction explaining the structure of the research We have included the following information on line 91 of page 5. “Hyeong Jun Ahn, as the first author, spearheads the study's conceptualization, protocol development, and manuscript drafting, collaborating with co-authors to refine the submission. Additionally, Minhee Kim provides valuable clinical expertise through an exhaustive literature review, synthesizing existing knowledge to inform the study's background and rationale. Furthermore, Kyle Ishikawa plays a pivotal role in the research endeavor, spearheading all aspects of data management, organization, and analysis, ensuring the integrity and accuracy of the study's findings.” 3- The working algorithm is not clear at all. In addition, the reasons for choosing them or the difference between them were not explained. Also, please pay attention to the numbering of titles Details and strengths about each algorithm have been added to the methods section where the finalized hyperparameters are discussed. Also, a flowchart of the general ML steps for each algorithm is presented as Figure 2. 4-The results for the metrics are presented in a table rather than in text form In consideration of the reviewer's valuable feedback, the decision was made to remove redundant metrics mentioned in the text, as it was deemed that the performance table offers a more precise presentation of the data. 5- The discussion is devoid of any mention of the results with their numerical values. It is also preferable to add the conclusions in a separate section As the reviewer pointed out, we have incorporated a summary of important numeric values (performance of each machine learning algorithm) from the results into the discussion section as follows (page 23, line 439). Our study reveals distinctive characteristics in the predictive performance of each machine learning model for metabolic phenotypes, as evident from their performance metrics. Both the Multinomial Logistic Regression and Neural Network models demonstrate moderate overall performance (AUROC of 0.818 and 0.814, respectively), with the former tending towards a balanced precision and recall, while the latter tends to prioritize precision over recall. In contrast, both the Random Forest and Boosted Trees methods exhibit competitive AUROC scores, underscoring their ability to capture underlying data patterns effectively. However, while the Random Forest model demonstrates higher precision, Boosted Trees display a more balanced precision-recall trade-off. Conversely, the K Nearest Neighbors model struggles with lower recall, indicating limitations in effectively identifying positive cases. Lastly, the Lasso Regression model, while maintaining decent precision, falls short in recall, suggesting potential challenges in capturing all instances of positive metabolic phenotypes. 6-Why is deep learning not mentioned as future work? Is it unlikely for certain reasons? Thank you for mentioning this. Our final model is a multilayer perceptron (MLP) which is a feed-forward neural network. Future work can be done to compare different neural net architectures and their suitability for our data. 7-References need to be more standardized and organized Thank you for providing feedback. We have carefully reviewed the references and standardized them to comply with the Vancouver style Reviewer #2: 1. It only used data from a single study/dataset (NHANES), limiting generalizability. Using multiple datasets could strengthen findings. Thank you for your advice. We added your feedback as limitation (page 26, line 502) 2. Performance comparisons were limited to 5 commonly used algorithms. Testing more recent deep learning methods may yield better results. Thank you for this suggestion. Future work can include more complicated neural net architectures that will address our dataset better. 3. Validating models on held-out test data from a later NHANES wave would better reflect real-world use, versus internal validation. We agree with and appreciate the reviewer's comments. As pointed out, using later NHANES data for validation could indeed enhance the value of the validation process. However, since NHANES data is collected in cycles with varying parameters, the parameters used in this study between 2007 and 2012 are not consistently collected in subsequent cycles, leading to a mix of available and unavailable parameters. This presents an inherent limitation in conducting validation with later cycles. Therefore, we believe it is reasonable to mention this as a limitation of our study (page 26, line 498) Correction in Discussion "Another limitation of our study is that although using NHANES data from different cycles for validation could potentially enhance the analytical value, a significant number of parameters used in our research were not included in data from other cycles. This discrepancy limited our ability to conduct a comprehensive validation." 4. Limited demographic/clinical covariates were considered; including more exhaustive potential predictors could impact results. We appreciate the reviewer's comment. We initially believed that applying as many parameters as possible (exhaustive potential predictors) to our analysis model would aid in improving outcomes across various machine learning scenarios. However, the actual addition of these parameters did not yield a significant improvement in results. This lack of significant enhancement may be attributed to the fact that in our study, we selectively applied parameters already associated with metabolic phenotype. 5. Interactions and non-linear effects between variables were not modeled, which may be important for complex phenotypes. Thank you for the suggestion. We have added a lasso regression model to the project for variable selection of the interaction terms. In the preprocessing phase for the lasso model, every two- and three-way combination of interaction terms were created for the lasso model to reduce. 6. Describe dataset features in more details and its total size and size of (train/test) as a table. Thank you for the comment. An overall summary table for our complete dataset has been added as Table 3, and details about the training-test split has been added to the flowchart (next item). 7. Pseudocode / Flowchart and algorithm steps need to be inserted. A flowchart for the general ML steps has been added to the end of the methods section. 8. Time spent need to be measured in the experimental results. Time to train the tuning models was added to Table 2. 9. Limitation Section need to be inserted. Limitation was inserted in line 489-513 at page 26. 10. The architecture of the proposed model must be provided The architecture of the final neural net has been described in the methods section and Figure 2 was created to show the architecture. 11. Address the accuracy/improvement percentages in the abstract and in the conclusion sections, as well as the significance of these results. Since we focus on the discriminating power we’ve emphasized AUROC in the abstract. AUROC quantifies the model's ability to discriminate between the positive and negative classes across all possible thresholds. It represents the probability that the model will rank a randomly chosen positive instance higher than a randomly chosen negative instance. 12. The authors need to make a clear proofread to avoid grammatical mistakes and typo errors. Thank you for your comments. We made a proofread again. 13. Add future work in last section (conclusion) (if any) We added a sentence about deep learning with more complex architectures in the final paragraph (page 27, line 519). 14. To improve the Related Work and Introduction sections authors are recommended to review this highly related research work paper: We have incorporated the content of the papers suggested by the reviewer into the introduction, relating to the expansion of machine learning applications within the healthcare sector (page 4, line 58). a) Optimizing epileptic seizure recognition performance with feature scaling and dropout layers b) Optimizing classification of diseases through language model analysis of symptoms c) Predicting female pelvic tilt and lumbar angle using machine learning in case of urinary incontinence and sexual dysfunction d) Utilizing convolutional neural networks to classify monkeypox skin lesions e) Hepatitis C Virus prediction based on machine learning framework: a real-world case study in Egypt Reviewer #3: Thank you for the hard work presented. It is nicely written and covers an important topic. Authors could take the advantages of ML algorithms in modeling one of the medical diagnosis applications. However, could provide them with some few comments: 1- Generally, along the sections, some paragraphs with the same information have been duplicated. Try to remove them as much as possible. Thank you for the feedback. As we addressed Reviewer 1’s feedback, we decided to remove redundant metrics mentioned in the text, as it was deemed that the performance table offers a more precise presentation of the data. 2- Authors could provide a section describing the dataset, its complexity (if) and other details that have direct impact on the outcome of the ML models. We addressed the complexity of NHANES data (page 25, line 473) and the impact of ML models (page 24, line 439). 3- The paragraph from the line 150 - 166, perhaps this paragraph be moved earlier when authors talk about preprocessing and Table 1. Thank you for this suggestion. We agree, and this paragraph has been combined with the paragraph before Table 1. 4- in line 152, there is Table X, is it the Table 1 or another table not being included in the document? Thanks for catching this. We removed this portion of the paragraph due to combining it with the paragraph before Table 1. 5- Why did not the authors provide a Figure shows the AUROC? While figures can be useful for visualizing certain aspects of data, such as trends or patterns, we believed that a table format would offer a more concise and straightforward presentation of the AUROC and other accuray values for each model. By organizing the outcome values in a table, readers can readily compare the performance of the different models across multiple metrics in a structured and easily interpretable format. This approach not only streamlines the presentation of results but also allows for a more efficient utilization of space within the manuscript. Furthermore, tables are particularly advantageous when presenting numerical data with a high degree of precision, as they provide a clear delineation of values without the potential distortion that can occur in graphical representations. By prioritizing clarity and precision in the presentation of AUROC and other accuracy values, we aimed to ensure that readers could accurately assess the performance of each model and make informed comparisons between them. 6- Authors could compared their results with some state-of-art works? We acknowledge the importance of contextualizing our findings within the broader landscape of existing research and it's essential to highlight that the comparison of our results with state-of-the-art works in the field of metabolic phenotypes for thyroid function was not feasible due to the lack of directly comparable studies in the current literature. The unique focus of our investigation on metabolic phenotypes specifically related to thyroid function presents a novel contribution to the scientific discourse, with limited existing studies directly addressing this intersection. 7- lines 258 - 260: this is a general advantage of ML that already being presented in the Introduction. Have you developed a real-time ML model? We have not created an interface where researchers can input patient characteristics. However, in our ML pipeline, we can produce outcome predictions if we have a structured dataset of predictors. Perhaps future work can look into creating a web dashboard for researchers to use. 8- If you believe that the dataset you are using is complex, why you used a regular Neural networks? why you did not implemented the deep learning? The single-hidden layer neural net did not perform better than the logistic regression, so we did not feel like we needed to explore more complicated architectures. However, that could be a point of future work. 9- How can we reflect the work of MacNell on yours? How can you use their conclusions to support your results? Actually we tried the sampling weights on our models but we could not find a significant difference. We believe that the impact will be present when outcome variables are significantly associated with the sampling weights. 10- You have mentioned about developing a confusion matrix. Can you show a Figure visualizing them? Yes, thanks for catching that. A confusion matrix for each model has been added to the results section after the performance table. https://doi.org/10.1371/journal.pone.0304785.r002
20 May 2024 Decision Letter - Vijayalakshmi Kakulapati, Editor Exploring the diagnostic performance of machine learning in prediction of metabolic phenotypes focusing on thyroid function PONE-D-23-34518R1 Dear Dr. Ahn, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. If you have any questions relating to publication charges, please contact our Author Billing department directly at authorbilling@plos.org. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Vijayalakshmi Kakulapati, Ph.D Academic Editor PLOS ONE Comments to the Author Reviewer #1: 1- The abstract needs a general beginning before going into the details of the proposed work. 2- Adding a paragraph at the end of the introduction that explains the structure of the research presented in all its sections 3- What does this mean on page 6: (supplemental table ). 4-The proposed model is not clear, and it is preferable to add a general structure or flow chart that explains the stages of work 5-Add at least two references published in the year 2024 Reviewer #2: An updated manuscript addressing previous comments and suggestions was evaluated positively. The updated submission demonstrates significant improvement and provides valuable insights relevant to the research community. Reviewer #3: The Authors have addressed the comments. In addition, within the comments of other reviewers, now the manuscript becomes more conducted and deserve to go for publication. ******** https://doi.org/10.1371/journal.pone.0304785.r003
Formally Accepted
30 May 2024 Acceptance Letter - Vijayalakshmi Kakulapati, Editor PONE-D-23-34518R1 PLOS ONE Dear Dr. Ahn, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset If revisions are needed, the production department will contact you directly to resolve them. If no revisions are needed, you will receive an email when the publication date has been set. At this time, we do not offer pre-publication proofs to authors during production of the accepted work. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few weeks to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. If we can help with anything else, please email us at customercare@plos.org. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Dr. Vijayalakshmi Kakulapati Academic Editor PLOS ONE https://doi.org/10.1371/journal.pone.0304785.r004

Open letter on the publication of peer review reports

PLOS recognizes the benefits of transparency in the peer review process. Therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. Reviewers remain anonymous, unless they choose to reveal their names.

We encourage other journals to join us in this initiative. We hope that our action inspires the community, including researchers, research funders, and research institutions, to recognize the benefits of published peer review reports for all parts of the research system.

Learn more at ASAPbio .