An Ensemble Machine Learning Technique for Detection of Abnormalities in Knee Movement Sustainability
Round 1
Reviewer 1 Report
The article is missing many bases and the specific changes have been attached in the pdf file below.
Comments for author File: 
Comments.pdf
Author Response
Please find the response in the attached file.
Author Response File: Author Response.pdf
Reviewer 2 Report
Line No.7: In abstract, “…. resulting in an imbalance in the obtained sEMG signal data”. It is purely subjective. Any deficit in the nerve conduction might results in imbalance or any muscle sprain might result imbalance. How do you conclude the imbalance sEMG signal data?
Line No. 56: Citation is missing
Line No. 71: Not clear
From Table 2, justify how the sEMG sensor points are consider as features?
Section 2.2.1 The result analysis for variation in the statistical values are not discussed. Author has superficially mention the variation but no proper justification for the variation across various points discussed
Author has illustrated the point of signal acquisition in figure 1 (6 sEMG data extraction points) but in the analysis, they have considered only 5 sEMG points (RF, BF, VM, ST, FX) and consider 4 channel, how? and why?
In pictures of Figure 3. legends and units are missing.
Table 5 is not being discussed in the paper.
Line 375: Confusion between CNN and NCC?
Table 6: Authors confused with metrics with models?
The title of table 6 is not correct.
Line 206: In this part…. presentation needs to be improved
Presentation language needs to be improved. Entire manuscript need to be revised for language.
Author has failed to focus on the novelty. Generally, performance of any model with Anomaly detection is better than without Anomaly. This could not be considered as an novel idea.
Author has failed to justify the abnormalities due to Knee movement sustainability exclusively compare to any other muscular deficiency or any other physiological deficiency
Author Response
Please find the response in the attached file.
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper presented abnormality identification in movements from imbalanced sEMG data and classified six activities (three normal and three abnormal) for standing, sitting, and gait positions. Different prediction classifiers are tested, and various anomaly removal techniques are discussed. The abnormality identification approach is interesting and effective. However, before the paper can be published, the following comments could be used to enhance the quality of the manuscript further.
1) In the Introduction section, the authors investigated the mobile sensing techniques and claimed that the EMG sensors could recognize motions in advance. However, from the reviewer's viewpoint, the nervous signals such as EMG, EEG, etc., and other musculoskeletal signals are purely biological reflections of the subject's explicit movement intention, which cannot be viewed as signals prior to movements. See for details: "From sensing to control of lower limb exoskeleton: a systematic review," Annual Reviews in Control, 2022, 53,83-96. In addition, richer exploration of state-of-the-art gait detection algorithms is suggested to expand the depth of the literature survey.
2) A more detailed data collection process should be added. For example, how the sEMG sensors are placed on the lower limb and what is the experimental procedure to ensure the robustness of data collection, etc.
3) Why did the authors apply band-pass filters ranging from 20Hz to 460Hz during data collection? This differs from line 113 on Page 5: "The data is pre-filtered with a band-pass filter that ranges from 25 Hz to 465 Hz". The authors needed to explain why double filters are applied to the data.
4) In Table 6, the authors are suggested to bold the best results for better readability.
5) The font sizes of some figures are suggested to be increased.
6) The authors should add some comparison experiments to demonstrate the advantages of the proposed approach.
7) Some typeset errors exist in the manuscript, such as line 56 on Page 2.
Author Response
Please refer to the attached file for the author's response.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Authors have failed to justify the novel model/ concept developed. It is an different application of existing technology
Author Response
Please refer to the attachment for the author's response.
Reviewer 3 Report
The author misunderstands sEMG signals and has not made a careful revision to the first comment in the previous round. In addition, the introduction section does not have an in-depth discussion of the current gait recognition techniques at all. More importantly, It is completely inconceivable that the introduction section does not have even one article for the past two years. Therefore, the author must polish the introduction section further to meet the publication requirements.
Author Response
Please refer to the attachment for the author's response.
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
Authors unable to express the novelty of the paper clearly .
Testing the data with existing model can not be considered as novel idea
Reviewer 3 Report
The revisions are fine.
