Dear Dr. Deatsch,

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 Apr 25 2025 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.

• 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,

Esedullah Akaras

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 note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, all author-generated code must be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.

3. We note that you have indicated that there are restrictions to data sharing for this study. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

Before we proceed with your manuscript, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., a Research Ethics Committee or Institutional Review Board, etc.). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of recommended repositories, please see

https://journals.plos.org/plosone/s/recommended-repositories. You also have the option of uploading the data as Supporting Information files, but we would recommend depositing data directly to a data repository if possible.

We will update your Data Availability statement on your behalf to reflect the information you provide.

4. Your ethics statement should only appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please ensure that your ethics statement is included in your manuscript, as the ethics statement entered into the online submission form will not be published alongside your manuscript.

[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?

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

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

Reviewer #1: I congratulate the authors for their work. Considering that the decrease in walking speed is an indicator of mortality for elderly individuals with chronic diseases, your work is very valuable.

However, I have a few suggestions and questions.

The intended use of deep learning and NN in elderly individuals with chronic diseases should be detailed in the introduction section.

The fact that the walking speed of the individuals was not measured with an objective equipment can be stated as a limitation.

Only the right side was used for grip strength. Domimancy was not taken into concern?

Measurements of participants' walking speed over time may vary because of the individuals performing the measurements. it should be stated how this standardisation was achieved. Was interrater reliability performed? ICC value ?

In your study, age, grip strength and BMI were reported as major predictors similar to the results in the literature. LR and NN results are also similar. In this case, the superiority of NN or the reason for its preference is insufficient. This section needs to be supported by the authors.

Reviewer #2: This study addresses a highly relevant topic in the field, contributing valuable insights that enhance our understanding of gait and its determinants on aging. The authors have tackled an important question with a well-structured approach, and their findings have the potential to inform both clinical practice and future research. The thoroughness of the methodology and the depth of analysis further strengthen the study’s significance, making it a noteworthy addition to the existing literature. Nevertheless, I believe that implementing the suggested revisions will further enhance the quality and impact of the study, strengthening its contribution to the field.

Intro

1. The authors highlight the need for larger datasets in deep learning analyses and provide sample sizes from previous studies (108, 239, 746, 1901) as examples. However, no references are provided to support these figures. Including citations for these studies would enhance transparency and allow readers to verify the source of this information. (Line 81-86)

2. The introduction would benefit from a clearer explanation of how the present study differs from previous research and which specific gap in the literature it aims to address. For instance, the statement 'Several attempts have been made using statistical models to predict gait speed changes from a narrow set of potential predictors' suggests prior work in this area. However, it would be helpful if the authors explicitly outlined how their study expands upon or improves these past efforts. Additionally, specifying the professional groups (health workers, physiotherapist, physician etc.) that could benefit from these findings would provide readers with a clearer understanding of the study’s relevance and impact.

Methods

3. The primary aim of this study appears to focus on the relationship between gait speed, aging, and mortality. However, gait speed can be influenced by a wide range of factors, including orthopedic, neurological, and other medical conditions. Wouldn't it be more informative to stratify the analysis by different subgroups to account for these variations? Discussing the potential impact of such factors and whether a subgroup analysis could enhance the findings would strengthen the study's interpretation and applicability.

4. I acknowledge that biostatistics and methodological details can sometimes be complex to fully follow, so I apologize if I have missed anything. That being said, I would like to raise a few points regarding the chosen cut-points and the study population. The authors have established clinically relevant cut-points for gait speed (0.8 m/s and 1.0 m/s) based on prior literature. However, gait speed thresholds may vary across different populations, particularly in individuals with neurological conditions such as Parkinson’s disease or stroke. Were disease-specific variations in cut-points considered when applying these thresholds to the study population? If not, this could be a limitation, as the same cut-points may not be appropriate for all individuals.

5. Additionally, the study utilized gait speed data from the BLSA cohort, where measurement intervals varied by age. Since older individuals had more frequent assessments, could this have biased the long-term predictive model? Were adjustments made to account for the potential overrepresentation of gait speed decline in older participants? A discussion on how these factors may have influenced the results would strengthen the study’s findings.

6. It would be better for the readers if the authors explained a point about the measurement and use of grip strength in their analysis. In the table describing input variables, grip strength is reported in kilograms (Table 2). However, it is specifically labeled as "hand grip muscles right (kg)." Could you clarify whether only the right hand was measured and analyzed? If so, what was the rationale for not including the left hand or using an alternative approach such as the mean of both hands or the dominant hand? Additionally, considering that grip strength was identified as a significant factor in the Sobol index analysis and highlighted as a basis for therapeutic intervention, do you believe that using only the right hand impacts the generalizability of your findings? If left-hand data were available, would incorporating it alter your results? Clarifying these points would strengthen the methodological transparency and the applicability of your findings.

7. Previous studies (e.g., Sadeghi et al., 2000) suggest that lower limb dominance plays a role in gait biomechanics, with the dominant limb often contributing more to propulsion while the non-dominant limb provides stability. Considering this, do you think assessing lower limb dominance could enhance the interpretation of gait speed determinants in your study?

Discussion

8. The discussion would benefit from a more detailed justification of why grip strength was emphasized as a key predictor of gait speed rather than more directly related lower-limb strength measures. While grip strength has been associated with overall strength and function, gait speed is likely more directly influenced by lower-limb muscle strength. Addressing this distinction and discussing potential reasons for the focus on grip strength over leg strength measures would provide greater clarity for readers.

9. The authors have considered multiple gait-related parameters in their predictive model. However, in clinical and health-related research, patient-reported outcome measures (PROMs) are frequently used to capture the patient's perspective. PROMs can provide valuable insight into how individuals perceive their mobility, fatigue, pain, or fear of falling, which are factors that could influence gait patterns over time. Did the authors consider incorporating PROMs while examining the determinants of slow gait? If not, this could be a potential limitation of the study, as relying solely on objective gait parameters may overlook important subjective experiences that contribute to mobility decline. Including PROMs in future research could enhance the model's generalizability and provide a more comprehensive understanding of aging-related gait changes.

Reviewer #3: This manuscript presents a valuable contribution to the field of aging and mobility research by exploring predictive modeling techniques for slow gait, a key biomarker of health and longevity. By leveraging data from the Baltimore Longitudinal Study of Aging (BLSA), the study compares the performance of a deep learning neural network (NN) with traditional logistic regression (LR) models in predicting current and future slow gait at different timeframes (6-year and 10-year). Additionally, the study identifies key determinants of gait decline, such as age, BMI, sleep quality, and grip strength.

The study is well-motivated and methodologically rigorous, making a compelling case for integrating machine learning into aging research. However, some methodological and analytical aspects require further clarification or justification:

Strengths________________________________________________

One of the key strengths of this study is its innovative application of deep learning to predict aging-related slow gait. While gait speed has been extensively studied as a predictor of health outcomes, the use of a neural network represents a novel approach that could potentially capture complex, nonlinear relationships between predictors and mobility decline. Additionally, by benchmarking the NN against logistic regression, the authors provide a robust comparative analysis, which strengthens the validity of their results.

The study also benefits from its use of a well-established longitudinal dataset (BLSA), which enhances the reliability of the findings. The long-term follow-up (6 and 10 years) is particularly valuable, as it allows for a more comprehensive understanding of mobility decline over time. Few studies have attempted to predict future slow gait over such an extended period, making this study particularly relevant for aging research.

Another commendable aspect of the study is its attention to data imbalance, a common issue in clinical datasets. By applying various class balancing techniques (RUS, SMOTE, SMOTE-ENN), the authors effectively address the skewed distribution of slow versus normative walkers. This methodological rigor enhances the study’s robustness and provides useful insights into best practices for handling imbalanced datasets in clinical prediction models.

Finally, the study has clear clinical relevance, as it focuses on clinically meaningful gait speed cut-points (0.8 m/s and 1.0 m/s). These thresholds align with established research on mobility disability and mortality risk, ensuring that the findings are directly applicable to clinical decision-making. The identification of modifiable risk factors (e.g., BMI, grip strength, sleep quality) further underscores the study’s potential impact, as these variables could inform targeted interventions for preventing mobility decline.

Areas for improvement and suggested refinements (Minor Revisions)_______________________

While this study presents a strong and well-motivated analysis of aging-related slow gait prediction using deep learning and logistic regression, there are some methodological and conceptual aspects that would benefit from clarification and refinement. These do not require major changes to the core analysis but would enhance the transparency, interpretability, and generalizability of the findings.

- One of the key strengths of this study is its comparison between neural networks (NN) and logistic regression (LR). However, the results indicate that NN performs only marginally better than LR, raising the question of whether the added model complexity is necessary. While the authors state that NN allows for capturing nonlinear relationships, there is no strong evidence that such relationships exist in this dataset.

Suggested Improvement: A brief discussion on whether nonlinear interactions between predictors were observed (or theoretically expected) would strengthen the justification for using NN. If applicable, referencing studies that have successfully demonstrated nonlinear patterns in similar aging-related predictions would provide useful context.

- Deep learning models are susceptible to overfitting, especially when applied to datasets with a relatively small sample size (1,363 participants). The manuscript mentions the use of dropout layers, but there is no explicit discussion of other overfitting mitigation strategies, such as hyperparameter tuning, cross-validation, or external validation.

Suggested Improvement: A short statement on whether techniques such as k-fold cross-validation or regularization methods were used would provide confidence in the model’s generalizability. If external validation was considered but not performed, a mention of this as a future step would clarify the scope of the current analysis.

- A common challenge with deep learning models is their black-box nature, which limits clinical interpretability. While the study employs Sobol sensitivity analysis to rank predictor importance, this approach does not fully address the need for clinically meaningful explanations of how individual variables contribute to predictions.

Suggested Improvement: A brief comparison between the Sobol index results and logistic regression coefficients would help readers understand whether the NN model identifies similar key predictors as traditional methods. Additionally, a sentence or two acknowledging the potential use of SHAP (Shapley Additive Explanations) or LIME (Local Interpretable Model-agnostic Explanations) in future research would be beneficial.

- The study replaces missing values with the median, but the rationale for this choice is not discussed. This is particularly relevant given that 15% of MMSE scores (cognitive impairment) were missing, which could introduce bias.

Suggested Improvement: A brief justification for using median imputation over other common methods (e.g., mean imputation, multiple imputation) would improve transparency. If a sensitivity analysis was conducted to assess whether the imputation method affected results, mentioning this would be valuable.

- The study is based on the BLSA cohort, which consists primarily of highly educated volunteers from a limited geographic region. This raises concerns about generalizability to more diverse populations, particularly individuals from different socioeconomic, racial, and educational backgrounds.

Suggested Improvement: The Discussion or Limitations section should briefly acknowledge this limitation and suggest future validation in more heterogeneous cohorts. Even a short statement recognizing the potential biases of volunteer-based longitudinal studies would enhance transparency.

- While the results clearly compare NN and LR, a concise summary statement highlighting the key takeaways—whether NN significantly outperformed LR or if the results were comparable—would help readers quickly grasp the implications.

Suggested Improvement: A small summary table comparing the strengths and weaknesses of both models in terms of accuracy, interpretability, and clinical applicability would make this section more accessible.

Specific Comments

1. Introduction

- Line 50-55: The claim that “gait speed is an essential predictor of overall health and well-being” is well-supported, but additional references on the impact of gait speed on cognitive decline would strengthen the argument.

- Line 72-75: The statement that NNs allow for capturing nonlinear relationships should be supported with examples from prior research in aging.

2. Methods

- Line 160-162: The handling of missing MMSE data (15%) is a potential limitation. Was a sensitivity analysis conducted to assess the impact of missing data on results?

- Line 231-234: The manuscript mentions using different solvers for logistic regression. Was feature selection or regularization applied to avoid overfitting?

3. Results

- Table 3: The dataset sizes after class balancing should be contextualized. How does this compare to the original dataset?

4. Discussion

- Line 357-363: The claim that NNs are advantageous for handling images and longitudinal data is valid but not demonstrated in this study. Consider clarifying that this is a potential future direction.

- Line 472-475: The discussion on modifiable risk factors (BMI, grip strength) is strong but could benefit from practical implications for clinical interventions.

5. Conclusion

- Line 607-612: The conclusion suggests that deep learning should be further explored, but it should acknowledge that logistic regression performed similarly, questioning the necessity of NN in this context.

**********

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: No

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.

Attachments

Attachment

Submitted filename: Review Form.docx

Prediction of future aging-related slow gait and its determinants with deep learning and logistic regression

PONE-D-24-56052R1

Dear Dr. Deatsch,

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,

Esedullah Akaras

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #2: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #2: Yes

Reviewer #3: Yes

**********

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

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #2: No

Reviewer #3: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #2: Yes

Reviewer #3: Yes

**********

Reviewer #2: I would like to sincerely thank the authors for their comprehensive and thoughtful responses to the reviewer comments. It is evident that considerable effort has been made to address the concerns raised during the initial review. The revised manuscript demonstrates substantial improvement, both in terms of clarity and scientific rigor. The additional explanations, methodological clarifications, and textual revisions have significantly strengthened the work’s contribution to the field. In particular, the enhancements to the Introduction and Limitations sections, as well as the added justifications regarding model choices and variable selection, are commendable. The manuscript is now more robust, transparent, and accessible to a broader clinical and academic audience. I appreciate the authors' diligence and responsiveness throughout the revision process.

Reviewer #3: All changes made by the authors are adequate and addressed the concerns of the reviewer. Therefore, I consider that the paper, in its current version, meets the necessary requirements to be published in Plos One.

**********

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 #2: No

Reviewer #3: Yes:  Eduardo Carballeira

**********