Preliminary Exploration of a Gait Alteration Index to Detect Abnormal Walking Through a RGB-D Camera and Human Pose Estimation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 The authors propose the Gait Alteration Index (GAI), which is an interpretable, continuous index designed to quantify deviations from typical gait patterns using markerless RGB-D acquisition and human pose estimation. Unlike approaches that directly regress ordinal clinical scores, the GAI conceptualizes gait impairment as a functional deviation, independently of specific diagnoses. A key contribution is the decomposition of gait into three clinically meaningful domains: 1) spatio-temporal characteristics, 2) dynamic stability, and 3) arm swing. These three domain-specific sub-indices result in an overall index. Methodologically, the work combines feature selection from a simulated gait dataset with centroid-based geometric modeling and validates the index against clinician-derived gait scores in a heterogeneous cohort including Parkinson’s disease, post-stroke, and non-pathological subjects. The paper’s primary novelty lies not in algorithmic complexity, but in offering a transparent, low-burden, and clinically interpretable aggregation layer that bridges raw kinematics and clinical judgment.
The methodology is appropriate, coherent, and largely robust for a preliminary validation study. Feature selection is carried out systematically using redundancy filtering, multiple selection algorithms, and nested cross-validation with hyperparameter tuning, which reduces overfitting and supports internal validity. The geometric formulation of the GAI is mathematically sound, scale-invariant (via standardization), and yields bounded, interpretable scores. The statistical analyses, is performed through non-parametric correlations, group-wise comparisons, and post-hoc testing, are suitable given the ordinal nature of clinical ratings and non-Gaussian distributions. Results are internally consistent: the strongest agreement emerges for the spatio-temporal domain, while reduced alignment for dynamic stability and arm swing is plausibly explained by both pose-estimation uncertainty and limitations of video-based clinical scoring. The conclusions are generally well supported by the data and are appropriately cautious, framing the GAI as a complementary indicator rather than a replacement for clinical assessment.
Several aspects could be strengthened to enhance clarity, generalizability, and clinical credibility:

    The authors claim to perform “low-cost” deployment. However this context should be contextualized relative to laboratory-grade motion capture, rather than framed in absolute terms.
    Second, the derivation of features from simulated gait data, while methodologically justified, remains a key limitation. Future studies should further validate them employing independent datasets or with pathology-specific centroids would be necessary to support broader generalization.
    Third, the assumption of equal weighting across the three sub-indices should be discussed more explicitly in relation to clinical practice, potentially supported by sensitivity analyses or clinician-informed weighting schemes. Maybe a weighted version of the GAI should be more clinically meaningful.
    Additional transparency would also be gained by clearly specifying the software environment used and by justifying methodological choices such as the use of Spearman correlation in explicit terms. The authors should state which software (e.g. R, Python, Matlab) used, their associated libraries / toolboxes / packages and their versions.
     Finally, although acknowledged by the authors, future work should prioritize multi-rater clinical annotation and longitudinal validation to assess reliability and sensitivity to change.

Addressing these points would substantially strengthen the manuscript’s translational impact without requiring major restructuring.

Author Response

Please see attached file. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This work proposed a Gait Alteration Index (GAI) to quantify gait abnormality as a functional deviation from typical walking patterns. GAI was analyzed and demonstrated rigorously. There was also a comparison between GAI and clinical scores. This is a good work. The following comments may be addressed before publication:

 

1. In section 1 line 64-69, the authors introduced the existing gait indices. This is good, but the information in this paragraph was limited. It'd be better to elaborate on at least one existing index, especially how it's data-demanding and difficult to calculate. As such, the readers would have a clear idea about the limits of the existing gait indices and how the proposed GAI is advantageous.
2. I liked the entire analysis for GAI. But the results could be further improved. Like what was mentioned in section 4.1, this work was limited by the case study. There could be more than one case study and bigger sample data. If not in this work, the authors should keep pursuing GAI in more field tests and validations.

Author Response

Please see attached file 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The article, titled "Preliminary exploration of a Gait Alteration Index to detect abnormal walking through a Red, Green, Blue, and Depth camera and Human Pose Estimation", captures an important research perspective for qualitative research, since it proposes an index designed to quantify gait abnormality as a functional deviation from typical walking patterns, independent of specific pathological diagnoses. According to the authors, the "Gait Alteration Index" represents a promising, low-cost, and interpretable indicator of gait abnormalities, suitable for objective assessment, screening, and longitudinal monitoring. Despite the interest and importance of the article, we cannot help but note serious textual errors that hinder the reader's ability to follow the content and analyze the research. We believe that some critical points, which better highlight the contents of the article, should definitely be re-examined by the authors. Initially, we would say that authors should limit themselves to the academic context of scientific publications. We would suggest a series of decisive changes in order to create a smooth introduction for the reader. These changes concern: 
1) Removing all abbreviations from the title and abstract, as these are the first approaches to the reader. Abbreviations can follow normally in the introduction. 
2) Reconsidering the functionality of the abstract as an introductory text and removing any information or data that can and should be presented later. Multiple and individual details are more likely to confuse readers than to inform them. 
3) As an introduction, they should conceive and reformulate an introductory text that will contain 1) the context of the problem, 2) the purpose, 3) the scientific and then the research questions, 4) the codified objective of the study, 5) the rough approach to the methodology, 6) the design tools and finally 7) the main findings and results and the contribution. All this with simple, concise, and focused sentences.                                                                                 4) At the end of the introduction, the authors should precisely redefine the scientific questions of the article, which will be answered in the conclusions.
5) After the scientific questions, the authors should list the individual chapters of the article with a connection to the questions and with a precise description, explanation, and justification.                                6) An immediately following and crucial issue concerns the handling of reports. Here, there are enough issues for a full review. Some references are together, for example, “[6-9], [10-12], etc. In our opinion, there is no point for the reader to approach references together. We do not use references simultaneously without explaining exactly what we are referring to! References in an article act as tools for documenting evidence. Do not allow confusion and allow the reader to interpret it “freely”. Rewrite and explain!!! Why do the authors refer to each reference anywhere in the article? What is the meaning of each reference? Explain why you use every other work in your documentation. Correct similar behavior everywhere in the paper! Explain why you refer to any reference! Furthermore, it is important not to repeat the same reference without explanation. If there is a reference that is repeated, the different approach should be explained; the repetition may be considered plagiarism.              7) Another important point is excessive paragraphing. Excessive paragraphing is not necessary, nor is an article a collection of notes. A minimum of three paragraphs per page or page section is sufficient. At least three paragraphs per page or page section are sufficient.                                                                                               8) Chapter 2 (2. Materials and Methods) is well done. There is a need for a pre-conception because the content is large, complex, and intricate. A short introduction might have clarified the direction of the manuscript, but it might also have been clarified by the restructuring of the first part. Let the authors judge this later, and after they have arranged the first chapter.                                          9) Avoid quoting parts without small text introductions. Note a small text between 3. Results and subsection 3.1. Phase 1 results. Apply it throughout the manuscript.                                                             10) Figure 3, Figure 4, etc. Explain how the specific images were produced, the tool, the accuracy, and generally the accuracy of what it represents. The content is not a given for everyone. It must be documented.                                                                                      11) Table 3, 4, 5, etc., please minimize the table captions and stick to what is necessary to describe the table in each caption. Describe the rest in the text.                                                                                  12) Figures 5, 6 - to 10.: It is impossible to accept these particular captions in any way, not even as an exaggeration. Explain the production tools and explain in detail what data was produced, which tools produced the charts, and what exactly they mean for the research findings.                                                                              13) The use of references in the discussion is not understandable. What purpose do they serve? Is the discussion not about the findings of the research? Are the findings not original?                      14) 5. Conclusions: to highlight the added value of the research. 

A comprehensive upgrade of the text and development on new foundations that we have noted is needed.

 

 

 

Author Response

Please see attached file. 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript is scientifically sound and presents a well-motivated, carefully executed preliminary study. The following minor revisions are requested to further improve clarity, presentation, and interpretability, without requiring any additional experiments or changes to the study design.

First, the authors are encouraged to slightly strengthen the conceptual explanation of the Gait Alteration Index (GAI) in the Methods section. While the mathematical formulation is clear, a short intuitive paragraph explaining how the three sub-domains (spatio-temporal, dynamic stability, and arm swing) jointly reflect functional gait deviation would improve accessibility for readers from non-technical clinical backgrounds.

Second, although the feature selection process is methodologically rigorous, the description is somewhat dense. The authors should consider streamlining the presentation of the feature selection pipeline, for example by briefly summarizing the multi-step selection logic at the beginning of the relevant subsection, to guide the reader before the detailed description.

Third, the Results section would benefit from clearer cross-referencing between figures, tables, and the corresponding textual interpretation. In particular, explicitly indicating which figures support each major observation would improve navigability and reduce cognitive load.

Fourth, in the Discussion section, the authors should more explicitly distinguish between limitations arising from pose-estimation accuracy and those inherent to clinical scoring scales. This distinction is already implicit in the text, but making it explicit would further strengthen the interpretative clarity of the findings.

Finally, a minor editorial pass is recommended to address small typographical issues and improve linguistic consistency throughout the manuscript.

Subject to these minor revisions, the manuscript is suitable for publication.

Author Response

Please see attached file. 

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Although the authors insisted on their original point of view on some issues, we must admit that they improved important parts of the article and altered the critical points of acceptance. A publication guide, such as guidelines from the University of Southern California, is not mandatory to bind all readers, nor does it mean that this is the only way to document results. Furthermore, what was mentioned as "a kind of plagiarism" is the constant citing of a source without explaining in detail why. 

Anyway, the authors

1) have added such a descriptive list of the Sections of the
manuscript, 

2) they improved the Introduction following our suggestions, 

3) they have shortened the abstract, following the guidelines for
Algorithms, reducing the word count below 200 words. They have retained numerical results, as we believe this provides, as per the journal’s guidelines, a summary of the article’s main findings by supporting them quantitatively,  

4) they explicitly specified the research questions of the paper that are later addressed in the discussion and the conclusions, supported by the results, and 

5) they have added such a descriptive list of the Sections of the manuscript. 

In the final version of the manuscript, the authors have further highlighted the relevance of their work in the current scientific landscape. Those changes are important and enough for publication. 

Author Response

Please see attached file.

Author Response File: Author Response.pdf