Intelligent Frozen Gait Monitoring Using Software-Defined Radio Frequency Sensing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper “Intelligent Frozen Gait Monitoring using Software Defined Radio Frequency Sensing” describes a non-invasive experimental setup for gait monitoring. The topic is well explained and thoroughly covered in the introduction and the related works sections. The authors present the methodology in a clear and accessible manner.

However, in my opinion, additional information could improve the understanding of the study's reproducibility. Specifically, the following points could be clarified:

• How many subjects were enrolled in the experiment? Some information about the study population might help to understand the results.
• Authors can add more information about the Testing procedures: How were the tests conducted? Did the participants follow a specific path? Were the durations of different activities consistent across trials? Were the algorithms fed observation windows or data from the entire trial?
• Other information about the Machine learning parameters can be introduced, for example, authors can indicate the hyperparameters used. 
  
• Can the obtained results be compared with other experiments found in the literature?

Adding these details would greatly enhance the transparency and replicability of the study.

Author Response

The responses to the reviewer comments are mentioned in the attached document.  

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This research provides a technically detailed explanation, clearly stating the objectives, methods, results, and significance of the study, with appropriate structure. However, some sentences are long and complex, making it difficult for readers to understand. Consider improving readability by simplifying the text and dividing paragraphs. Additionally, the following points should be addressed more clearly:

 

Abstract:

The results section of the study is persuasive, but it could be more effective with comparisons to other results and specific numerical data. The entire text is compiled into long paragraphs, making it challenging to read. Breaking it down into sections can help organize the structure and improve readability. The following points should be elaborated more clearly:

·        Introduction: Challenges of FOG and limitations of conventional methods

·        Research Objectives: Introduction of SDR and RF sensing

·        Methodology: Data collection, processing, and model development

·        Results: Improved accuracy and superiority of the deep learning model

·        Conclusion: Significance of the proposed system and future prospects

Conclusion:

The conclusion section should be emphasized more and include future prospects. For example: "The system developed in this study has shown great potential for accurate real-time monitoring of FOG and other PD symptoms. Future experiments in more diverse physical activities and environments are necessary to validate the system's versatility and reliability. Furthermore, developing user interfaces aimed at application in patients and medical settings is a crucial task."

Introduction:

This paper discusses a study on detecting "freezing of gait (FOG)" motor disorder in Parkinson's disease (PD). The text explains the background, issues, limitations of conventional methods, the new proposed method, the study's contribution, and the structure of the paper.

The challenges and limitations of conventional detection methods should be clarified further. For example, the specific issues with "wearable sensors" mentioned in the latter part of the text (discomfort, data overload, etc.). The need for a model like the one in this study should be discussed, including aspects of diagnostic accuracy and its relation to treatment, with literature citations.

Also, a description that derives hypotheses from the detection with SDR methods and the expected contributions is necessary. Additionally, by supplementing more specific advantages of "SDR technology," the reasons for choosing this technology should be made clear to the readers. Furthermore, emphasize the novelty of the study and expand on how the proposed methods (use of SDR and wireless technology) differ from conventional methods and improve upon them.

Method:

The description of AI methods is deemed sufficient. However, clarify the selection criteria for the patients during the study period. What were the exclusion criteria? The details about the subjects should be clearer.

Results: and discussion

The causal sequences and accuracy of various machine learning methods are understood. It is necessary to show the frequency of FOG in each task (1) to (5) mentioned in lines 267-268 within the methodology. It was found that the FOG patterns could be classified into six categories using machine learning and deep learning models. Explain the clinical significance of these six categories and what differences in FOG symptoms these six categories indicate. Additionally, clarify the significance of this method from the differences in detection compared to conventional FOG.

Author Response

The responses to the reviewer comments are mentioned in the attached document.  

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript introduces a method for gait recognition using Software-Defined Radio (SDR). While the topic is relevant and has potential applications in non-invasive biometrics, several critical aspects of the study require significant clarification and improvement to enhance its scientific rigor and practical applicability. Below are the main concerns:

 

Motivation and Experimental Justification

The manuscript lacks clear articulation of the motivations for the proposed method and how the experiments are designed to validate it. The rationale behind the experimental conditions and the choice of parameters is underexplored, leaving the reader questioning the foundational design principles.

 

Wearable Devices and User Discomfort

The paper claims that wearable devices cause discomfort to users, positioning SDR-based solutions as a preferable alternative. However, this assertion is insufficiently supported by evidence or argumentation. It is recommended to provide quantitative or qualitative data, or references to prior studies, that substantiate this claim.

 

Measurement Setup

The measurement setup appears overly controlled, limiting the generalizability of the findings. Key details, such as the distance between the transmitter and receiver, are missing. Moreover, the activity time is specified as 15 seconds, raising questions about how such a duration is feasible in the constrained experimental space. Clarification and justification of these aspects are essential.

 

Inconsistency in Data Representation

Figure 4 reports only 3750 samples, which correspond to approximately 2.5 seconds of activity, rather than the stated 15 seconds. This discrepancy undermines the credibility of the experimental results and requires immediate resolution. Additionally, the absence of measurement units throughout the manuscript makes it challenging to interpret the reported data accurately.

 

Overly Controlled Experimental Environment

The confusion matrix tables presented in the manuscript appear overly idealized, suggesting that the experiments may have been conducted in an excessively controlled environment. This raises concerns about the real-world applicability of the proposed method. A more realistic experimental setup, including noise, variability, and less constrained conditions, is strongly recommended.

 

Lack of Comparative Analysis

The manuscript does not provide a comprehensive comparison with existing state-of-the-art techniques for gait recognition. Including a comparative table summarizing the performance of the proposed method against similar approaches would significantly strengthen the study's contribution to the field.

 

Need for Improved Experimental Validation

The experimental section requires considerable refinement. Testing the method in a realistic environment with greater diversity in experimental conditions is necessary to validate its robustness and practical utility.

 

In conclusion, while the manuscript addresses an interesting topic with potential significance, its experimental design, analysis, and presentation require substantial improvements. Addressing the concerns outlined above will greatly enhance the clarity, credibility, and impact of the work.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

The responses to the reviewer comments are mentioned in the attached document.  

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The authors described revised AI and ML methods for more conveniently and efficiently detect freezing gait in Parkinson's patients. They made a reasonable case for this in their introduction and made a clear purpose statement. Their methods section will need some important additional information. The results were less clear because they failed to provide complementary text for figure 4 (which seemed to be a primary result). There was no discussion section that 1) briefly restated what problem they attempted to address, 2) the rationale/need for this study, 3) the primary purpose of the study, 4) a summary of their primary findings and a cogent discussion of these findings that are clinically meaningful, 5) a comparison of their findings to similar published research, etc. Their conclusion appeared to partly be a discussion. A couple of figures failed the stand alone test because I was not able to know what the figure was telling me from its caption and the absence of explanation in the narrative.

The authors may have missed some limitations. For example, their sample appeared to include only normal healthy individuals yet they were interested in freezing gait which appears in individuals with Parkinson's disease. 

There were several word choice and word tense errors, which I highlighted. This manuscript will need significant expert English grammar editing. I highlighted a few examples early in the manuscript but left the rest to the authors and a grammar consultant.

See pdf for specific comments.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Noted above.

Author Response

Response file is attached 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

This paper demonstrates that sensing with the SDR-RF system and utilizing a deep learning model enables highly accurate detection of freezing of gait in gait monitoring. Based on feedback from reviewers, parameters related to freezing of gait, patient attributes, and hyperparameter tuning of the deep learning model were also presented. A platform that leverages machine learning and deep learning algorithms is considered useful for predicting gait disorders and falls in future rehabilitation, and further advancements are expected.

Comments on the Quality of English Language

none

Author Response

Response file is attached.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

There were some improvements in this revision; however, significantly more work is needed. See comments in the pdf.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The quality of the English language is still insufficient.

Author Response

Reviewer response is attached. 

Author Response File: Author Response.pdf

Round 3

Reviewer 4 Report

Comments and Suggestions for Authors

I appreciate your patience. This revision shows improvement; however, there are still many word choice and word tense errors. The major deficiencies that need to be addressed include: additional information in the methods section which is vital if anyone would wish to replicate this work, a more detailed discussion about the practical application of the signals generated; and how the novel method has advantages over other current methods. See the pdf for more specific comments.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

This has gone through several revisions but too many grammatical, etc. errors were evident.

Author Response

Response file is attached

Author Response File: Author Response.pdf

Round 4

Reviewer 4 Report

Comments and Suggestions for Authors

This revision has improved; however, there are still some issues that are identified in the pdf. There were lingering word tense errors and word choice errors as well. I appreciate your patience and perseverance. 

Please in the next revision highlight what is different based on my comments. It seemed to me I found things this time that I commented on before.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

a few minor errors persist.

Author Response

Response file is attached.

Author Response File: Author Response.pdf