PONE-D-21-32222The Portiloop: a deep learning-based open science tool for closed-loop brain stimulationPLOS ONE

Dear Dr. Bouteiller,

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, revise manuscript for clarity, you are requested to do substantial restructuring and shortening of the manuscript. Please, clearly describe  the goal and methods of the study, and provide the details of the design and data analysis.

For datasharing, please, use a scientific repository (e.g. Figshare, Harvard Dataverse,...)

More than half the figures (Figures 1, 3-7, 9) only highlight processes and methods. The rest of the figures cannot be interpreted without the main text. Therefore, the scope of analyses and presentation of the results required significant improvement.

Consider concatenating figures to reduce the number of figures.

The potential for a toolbox is evident, but such a toolbox’s utility and functionality are not demonstrated, but rather implied. The prototypical toolbox distracts from the analyses.

How could someone get one of these devices? Are readers expected to be able to build them, following your instruction (in which case the instruction would have to be much more detailed)?

It would be nice to see a picture of the device and how it can be used in the lab environment.

Reduced ANNs for maximizing Recall and Precision are not developed or analyzed. No reduced ANN matched SpindleNet’s Recall (% true positives). Reduced ANNs all had superior Precision (% positives found). Why? Three reduced models strike a balance between Recall and Precision with f1 scores of 0.61, but no model variations achieved the performance of expert raters. Why? All the reduced models perform essentially the same.

Minor comments: Table 1. Please spell out IExp, as this term / abbreviation is not used in the text. No stimulation is actually conducted in this study. Do the Authors mean actuate?

Do not use a special symbol (infinity) in the title, this will be very difficult for indexing, search engines etc.

Please submit your revised manuscript by Feb 17 2022 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,

Gennady S. Cymbalyuk, Ph.D.

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. 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: Partly

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: N/A

**********

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

**********

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: No

Reviewer #2: 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: This is a very long and non-stringent, yet potentially very interesting manuscript. I am not able to give an in-depth review at this point, since reading the manuscript is rather confusing, and it would probably take me several days to figure out everything that you are trying to say. I would advise you to do some substantial restructuring and shortening work, then I'll be happy to provide an in-depth review. For the moment, I have some mostly technical remarks:

* I would not recommend to use a special symbol (infinity) in the title, this will be very difficult for indexing, search engines etc.

* Also, it is not the gold standard to share data via a private Google Drive, have you considered using a scientific repository (e.g. Figshare, Harvard Dataverse,...)

* If my computer is counting correctly, the word count is ~10,000, which is clearly too long for the manuscript to be read by most people, including myself.

* This manuscript should not come with 13 figures. Since most of them contain very little information, consider concatenating them to three or four figures with several subplots.

* It is unclear to me what exactly you are presenting in this paper - you say it's about a new device, but at the same time you talk a lot about machine learning and software benchmarking fundamentals, and sleep physiology, which in my humble opinion is just too much for one manuscript. Please try to be concise and have a clear focus for the manuscript. For example, the best ANN architecture for spindle detection is in principle independent of any hardware implementation, and could therefore be dealt with in a separate paper (or in the Supplementary Material, if you do not deem it worth of a separate paper).

* Related to this: You open-sourced the software, but how could I get one of these devices if I wanted to? I may have missed it, but I found no information on this. Was it self-built by your lab or did you have an industry partner (which I don't assume since there is on conflict of interest declared). Can I buy it from you (in which case a COI would need to be declared), or am I expected to build it myself, following your instruction (in which case the instruction would have to be much more detailed)? Without at least mentioning any prospective way that I could get hold of one of these devices in the future, I'm not sure how useful a paper describing it will be.

* by the way, it would be nice to see a picture of the device and how it can be used in the lab environment.

* Related to this, I did not find any information about data storage for offline analysis, and how it would integrate with existing lab infrastructures (e.g. Labstreaminglayer).

To sum up, I think closed-loop stimulation is a relevant field with a lack of well-functioning and easy-to-use devices. I would therefore highly appreciate if you could bring your paper into a more digestable shape, so that your work can receive its due appreciation.

Reviewer #2: Summary:

The Authors developed hardware that can implement Artificial Neural Networks (ANNs) for detecting sleep spindles using FPGAs. Design constraints were imposed to make the hardware lightweight and portable for potential application in close-loop stimulation studies. This study focuses on a deep-learning, open-source toolkit for model-driven design of the FPGA ANNs, Portiloop, using a public sleep dataset (MODA) for training and testing. Such a toolkit could be useful for studies on sleep and brain stimulation, but the results are lacking. I have three concerns.

Concern 1 – The potential for a toolbox is evident, but such a toolbox’s utility and functionality are not demonstrated, but rather implied. The prototypical toolbox distracts from the analyses.

Concern 2 – Reduced ANNs for maximizing Recall and Precision are not developed or analyzed. No reduced ANN matched SpindleNet’s Recall (% true positives). Reduced ANNs all had superior Precision (% positives found). Why? Three reduced models strike a balance between Recall and Precision with f1 scores of 0.61, but no model variations achieved the performance of expert raters. Why? All the reduced models perform essentially the same.

Concern 3 – More than half the figures (Figures 1, 3-7, 9) only highlight processes and methods. The rest of the figures cannot be interpreted without the main text. Therefore, the scope of analyses and presentation of the results required significant improvement.

Minor comments:

Table 1. Please spell out IExp, as this term / abbreviation is not used in the text.

No stimulation is actually conducted in this study. Do the Authors mean actuate?

**********

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: Yes: Marius Keute

Reviewer #2: 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.

PONE-D-21-32222R1The Portiloop: a deep learning-based open science tool for closed-loop brain stimulationPLOS ONE

Dear Dr. Bouteiller,

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, provide actual examples of the EEG and how Portiloop classifies events, in this case, sleep spindles within the time series, and how this classification compares to expert manual classification. 

Abstract. Please, define and mention MODA in the Abstract.

Please, consider removing Figures 3.  In Figure 6, it will help to see the raw and processed signal, along with its envelope, and this could be integrated into Figure 2.

Table 2. The absence of visuals to support Portiloop's performance is a concern. The only figure that shows Portiloop’s processed EEG are in Figure 2. This paper needs a figure showing how Portiloop's different variations detect spindles in EEG compared to the nominal ground truth, the MODA dataset. MODA vs. 2-input Portiloop vs. 2-3 variants of Portiloop that favor speed over classification accuracy.

Table 2. Too many abbreviations and undefined terms in Row Set 2 -- on-line detection. Figures and tables should be interpretable without the main text. Therefore, the table needs a legend to describe, at a high level, what is meant by 1- and 2-inputs, ablation, td, p1, so forth.

Figure 7. As is, Figure 7 is abstracted from real run / compute times. It would helpful to contrast the classification performance before and after optimization. What is the trade-off between compute time and classification with a high software and hardware cost (without PMBO), versus a low software and hardware cost (with PMBO). Show EEG output of two ANNs off and on the Pareto front.

Figure 8. Figure 8 needs to be rotated clockwise 90 degrees.

Figure 9. It's not clear what score and threshold mean in this figure without referencing Figure 2 and the text. Figures should be mostly interpretable in isolation. Please, show a panel (as in Fig. 2) with the EEG colored and annotated to show what are true positives, false positive, etc. for stimulation timestamps. It may also help to see an example of a delay relative to the spindle (and its duration).

Please, consider whether you could bring the main text word count closer to 4000 or 5000 words and push more technical details to the Supplement or the Github repo. 

Please submit your revised manuscript by Jun 11 2022 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,

Gennady S. Cymbalyuk, Ph.D.

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

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: (No Response)

Reviewer #2: (No Response)

**********

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

Reviewer #2: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: 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

Reviewer #2: 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: No

Reviewer #2: 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 authors have done a good job with the revision. Objectives are laid out much clearer, and a number of well-made figures help the reader, and there is a comment about how to obtain such a device. There is also a My main concern with the manuscript, however, has not been adressed - it is still clearly too long to be read by a majority of the scientific community. While a part of the Methods section has been shifted to the Supplement, the word count for the main text is still between 9000 and 10000. I think the chances of having people actually read, and cite, the article would be much higher if the authors could bring the main text word count closer to 4000 or 5000 words and push more technical details to the Supplement or the Github repo. Also, the authors should be aware that the target audience are probably experimental neuroscientists more than computer scientists or electrical engineers, so a bit more information about the example study (sleep spindles) and a bit less technical detail would probably be appreciated.

Reviewer #2: This design paper covers Portiloop, a closed-loop and lightweight tool for event classification in timeseries using deep learning. The paper uses detection of sleep spindles from a public dataset, MODA, as a first test case and proof of concept. The manuscript, while dense in text, adequately describes the motivation and methods for constructing Portiloop. However, the results are left mostly in text form. Nontechnical readers will benefit from seeing actual examples of the EEG and how Portiloop classifies events, in this case, sleep spindles within the time series, and how this classification compares to expert manual classification. Therefore, my remaining concerns for this manuscript are in its presentation of the results. Some figures show flow chart of methods adequately described with text alone, and some figures need to show real examples of classified EEG data. My specific comments are below:

Abstract. I think it will help to define and mention MODA in the Abstract to give it a search link to a term relevant to sleep researchers.

Figures 3 and 6. I agree with Reviewer 1’s previous concern of too many figures. Figures 3 and 6 appear unnecessary. However, in Figure 6, it will help to see the raw and processed signal, along with its envelope, and this could be integrated into Figure 2.

Table 2. The absence of visuals to support Portiloop's performance is a concern. The only figure that shows Portiloop’s processed EEG are in Figure 2. This paper needs a figure showing how Portiloop's different variations detect spindles in EEG compared to the nominal ground truth, the MODA dataset. MODA vs. 2-input Portiloop vs. 2-3 variants of Portiloop that favor speed over classification accuracy.

Table 2. Too many abbreviations and undefined terms in Row Set 2 -- on-line detection. Figures and tables should be interpretable without the main text. Therefore, the table needs a legend to describe, at a high level, what is meant by 1- and 2-inputs, ablation, td, p1, so forth.

Figure 7. As is, Figure 7 is abstracted from real run / compute times. I think it would helpful to contrast the classification performance before and after optimization. What is the trade-off between compute time and classification with a high software and hardware cost (without PMBO), versus a low software and hardware cost (with PMBO). Show EEG output of two ANNs off and on the Pareto front.

Figure 8. Figure 8 needs to be rotated clockwise 90 degrees.

Figure 9. It's not clear what score and threshold mean in this figure without referencing Figure 2 and the text. Figures should be mostly interpretable in isolation. I recommend showing a panel (as in Fig. 2) with the EEG colored and annotated to show what are true positives, false positive, etc. for stimulation timestamps. It may also help to see an example of a delay relative to the spindle (and its duration).

**********

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: Yes: Marius Keute

Reviewer #2: 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.

The Portiloop: a deep learning-based open science tool for closed-loop brain stimulation

PONE-D-21-32222R2

Dear Dr. Bouteiller,

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,

Gennady S. Cymbalyuk, Ph.D.

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

Reviewer #2: 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

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: 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

Reviewer #2: 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

Reviewer #2: 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: I still think it is quite long, but otherwise I have no further complaints. I recommend to accept this paper.

Reviewer #2: While I feel it would have helped to show non-technical readers the explicit performance and tradeoffs between the different ANNs (e.g., examples of spindle detection on time series with maximal recall vs. maximal precision vs. maximal f1 vs. maximal hardware costs vs. optimal tradeoff in software and hardware costs etc.), this is stylistic and not essential. Thank you for addressing my technical concerns, and I have no further comments.

**********

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: Yes: Marius Keute

Reviewer #2: No

**********