A deep learning backcasting approach to the electrolyte, metabolite, and acid-base parameters that predict risk in ICU patients

Albion Dervishi

doi:10.1371/journal.pone.0242878

Peer Review History

Original SubmissionMarch 30, 2020
20 Aug 2020 Decision Letter - Feng Luo, Editor PONE-D-20-09103 A Deep Learning Backcasting Approach to the Electrolyte, Metabolite, and Acid-Base Parameters That Predict Risk in ICU Patients PLOS ONE Dear Dr. Dervishi, 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 Oct 04 2020 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: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols We look forward to receiving your revised manuscript. Kind regards, Feng Luo Academic Editor PLOS ONE Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1. 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. To comply with PLOS ONE submission guidelines, in your Methods section, please provide additional information regarding your statistical analyses. For more information on PLOS ONE's expectations for statistical reporting, please see https://journals.plos.org/plosone/s/submission-guidelines.#loc-statistical-reporting. 3. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information. [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 ******** 2. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: 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 ****** 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 ****** 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: The manuscript titled, “A Deep Learning Backcasting ...” studies a deep learning model for prediction of risk using critical care data. The claim is that a ‘bakcasting’-based approach can increase the efficacy of clinical decision making using deep learning. The study is an addition to the already existing knowledge in AI based approaches utilizing clinical data. The manuscript is lacking in several aspects. I suggest the author revise the manuscript taking into account the comments and suggestions below. 1) The author has been rather frugal in providing explanations and detailing the study. Given the broad readership of the journal, the manuscript needs an overhaul. I would suggest the author to adhere to the standard writing style of Plos one. 2) Related to above, the Abstract needs to be written in a way that clearly states the motivation and introduces the problem, analyses techniques and discusses the main findings. The current format is not suitable. 3) The Introduction should be extended to provide a motivation for the study while discussing the previous research work, what has been lacking therein, and why the current study should be considered as an improvement. 4) Inside the methodology the author needs to elaborate on the technique of ‘backcasting’. Also the flow diagram, Fig. 1 should be discussed more. In general, the backcasting technique needs to be separately motivated. One would also like to understand, in brief, the usual practices of manually processing such data by medical personnel - how these particular numbers are useful in monitoring the health of a patient. I understand the above has been elaborated in the Discussion (referring to Table 3). But would be better to include in the methodology, at least some of it. The author has actually referred to Table 4 & 5 before Table 3. 5) In the Discussion and conclusion the author should again highlight the importance and shortcomings of the study, and be more concrete as to why “deep learning model in recognizing patients at risk for deterioration needs clinical validation”. Are there confounding factors? 6) Correlation aspects between the features have not been discussed. The author should perform more exploratory analysis of the dataset like PCA, etc. 7) “Our model uses multilayer neural networks, but a precise evaluation of their architecture would be beyond the scope of this study” - but a precise description of the model architecture/details and chosen parameters and hyper-parameters is a must. 8) The authors need to compare the deep learning method with some generic baseline model, or some less complex model like Naive Bayes or logistic regression. 9) Why not use something like permutation feature importance? 10) How does the age and gender of the patients (training set) influence the prediction? ****** 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 [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.0242878.r001
Revision 1
14 Oct 2020 Author Response Dear Editor, I wish to submit a revised manuscript " A Deep Learning Backcasting Approach to the Electrolyte, Metabolite, and Acid-Base Parameters That Predict Risk in ICU Patients.” for PLOS ONE. Manuscript Number: PONE-D-20-09103 I am very grateful to you and the reviewer team comments and thoughtful suggestions. Based on these comments and suggestions, I made careful revision to the original manuscript. Here are the responses to the reviewers’ comments and suggestions. The reviewer’s comment 1. The author has been rather frugal in providing explanations and detailing the study. Given the broad readership of the journal, the manuscript needs an overhaul. I would suggest the author to adhere to the standard writing style of Plos one. Response 1. We are very grateful to your comments on the manuscript. By this revision, additional complementary features were written. I wrote additional explanations of the methodology and discussion regarding the model also I tried to follow the standard writing style of Plos one. The reviewer’s comment 2. Related to above, the Abstract needs to be written in a way that clearly states the motivation and introduces the problem, analyses techniques and discusses the main findings. The current format is not suitable. Response 2. We thank you very much for your comments and advice. I rewrote the Abstract that states the main findings, results, or arguments of the project. The reviewer’s comment 3. The Introduction should be extended to provide a motivation for the study while discussing the previous research work, what has been lacking therein, and why the current study should be considered as an improvement. Response 3. Thank you for this suggestion. In the Introduction, we supplemented facts, references and detection time of critical laboratory values. However, based on our knowledge to date, building a model-based electrolyte, metabolite, and acid-base parameters of patients near the end of their ICU stays has not been performed. The reviewer’s comment 4. Inside the methodology the author needs to elaborate on the technique of ‘backcasting’. Also, the flow diagram, Fig. 1 should be discussed more. In general, the backcasting technique needs to be separately motivated. One would also like to understand, in brief, the usual practices of manually processing such data by medical personnel - how these particular numbers are useful in monitoring the health of a patient. I understand the above has been elaborated in the Discussion (referring to Table 3). But would be better to include in the methodology, at least some of it. The author has actually referred to Table 4 & 5 before Table 3. Response 4. Thank you for pointing this out. I tried to address better this issue. The backcasting technique is written separately in the Methods as “Backcasting approach algorithm “. In the Discussion, I discussed more, how I developed the backcasting model. Additionally, in the discussion, I described separately how this model can be addressed in the Backcasting approach "Backcasting approach for everyday application". For example, Adult patient admitted in the ICU with Ketoacidosis. The patient monitored every 6-8 Hours (sometimes hourly) with arterial blood gas analyze, where included all parameters in our Model. The patient at 8 a.m. was presented with laboratory values: Ph 7.12, Lactate 3.2, HCO3 10, Glucose 417, K 4.9, Na 136, Cl 99, Ca 10.5. Our model shows if untreated a risk of death within 0-8 hours is 99.59%. During treatment (fluids, insulin etc.) at 9 a.m. shows with laboratory values: Ph 7.21, Lactate 2.8, HCO3 14, Glucose 325, K 4.2, Na 135, Cl 98, Ca 10.4. calculated a risk of death from 65.72%. Conclusion: During 1 hour of treatment we have reduced the risk of Patient death from 99.59% to 65.72% based on Deep Learning LEMA-model with AUC of 0.982, specificity of 0.947, and sensitivity of 0.935. The R-based interactive web applications for risk adjustment in this link: https://albiondervishi.shinyapps.io/LEMA/ The reviewer’s comment 5. In the Discussion and conclusion, the author should again highlight the importance and shortcomings of the study and be more concrete as to why “deep learning model in recognizing patients at risk for deterioration needs clinical validation”. Are there confounding factors ? Response 5. Thank you for recommendation. I tried to address this issue throw the Manuscript. There are no confounding factors, I supposed it will be a while until the models can be widely used (e.g. APACHE II 1985, SOFA 1996), and still, we perform data analysis based on these models. The reviewer’s comment 6. Correlation aspects between the features have not been discussed. The author should perform more exploratory analysis of the dataset like PCA, etc. Response 6. Thank you for this hint. Indeed, it helped me to comprehend better the interaction between parameters. I used the standard Pearson’s correlation matrix. Additionally, I added also the maximum and minimum of my data in Tables 1 and 2. The reviewer’s comment 7. “Our model uses multilayer neural networks, but a precise evaluation of their architecture would be beyond the scope of this study” - but a precise description of the model architecture/details and chosen parameters and hyper-parameters is a must. Response 7. We thank you for advice. I described the model architecture, chosen parameters and hyper-parameters. Also, I design a Figure (Fig.1) regarding my deep learning model. The reviewer’s comment 8. The authors need to compare the deep learning method with some generic baseline model, or some less complex model like Naive Bayes or logistic regression. Response 8. We agree with this and have incorporated your suggestion throughout the manuscript. I compared Deep Learning with a generalized linear model (GLM) as logistic regression and Random Forest. An important conclusion came to a comparison between them. The reviewer’s comment 9. Why not use something like permutation feature importance? Response 9. We are very grateful to your reflective suggestions. Because I did a correlation together with models’ comparison, I believed it can be enough for parameter interaction and importance to the model. The reviewer’s comment 10. How does the age and gender of the patients (training set) influence the prediction? Response 10. Thank you for this question. I added in this revision that age and sex were included in the data selection but were not incorporated in the model for risk adjustment. I consider that age and sex are not physiological parameters and I assume that cannot contribute in my model. In conclusion, according to the comments, we attempted to revise the original manuscript. During the revised process, I added two Figures (Figures 1 and 4) also, added one table (Table 3). Hope the manuscript has been improved and it has reached the journal standard. Please feel free to address all correspondence concerning this manuscript to me at albiondervishi@gmail.com Sincerely, Albion Dervishi Attachments Attachment Submitted filename: Response to Reviewers.docx https://doi.org/10.1371/journal.pone.0242878.r002
11 Nov 2020 Decision Letter - Feng Luo, Editor A Deep Learning Backcasting Approach to the Electrolyte, Metabolite, and Acid-Base Parameters That Predict Risk in ICU Patients PONE-D-20-09103R1 Dear Dr. Dervishi, 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 for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, 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, Feng Luo Academic Editor PLOS ONE Additional Editor Comments (optional): Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation. Reviewer #1: All comments have been addressed ******** 2. 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 ****** 3. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes ****** 4. 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 ****** 5. 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 ****** 6. 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: The author has indeed addressed all the comments during the first round of the review. The manuscript now appears to be more detailed and rigorous. I would recommend publication. However, I would suggest making the Abstract more concise and less heavy with technical details. The author might look up other papers in Plos one and somewhat adopt the styles. ****** 7. 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 https://doi.org/10.1371/journal.pone.0242878.r003
Formally Accepted
8 Dec 2020 Acceptance Letter - Feng Luo, Editor PONE-D-20-09103R1 A Deep Learning Backcasting Approach to the Electrolyte, Metabolite, and Acid-Base Parameters That Predict Risk in ICU Patients Dear Dr. Dervishi: I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department. 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 plosone@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. Feng Luo Academic Editor PLOS ONE https://doi.org/10.1371/journal.pone.0242878.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 .