Novel Feature Extraction and Locomotion Mode Classification Using Intelligent Lower-Limb Prosthesis

Round 1

Reviewer 1 Report

This paper presents a locomotion mode classifier for lower-limb prosthetics using machine learning methods by using extracted features collected from six user data to establish its superior performance. As claimed by the authors, the main contributions are the proposal to use these features in this application, the description of the gait fluctuation with a Gaussian distribution, and the reported performance. However, the paper can be improved in various aspects.

1. Professional editing service is strongly encouraged. There are too many grammar issues, even errors, throughout the paper. To name a few as below, and there are more.

- Title: inconsistent caption

- Line 35. research or Research?

- Line 46: transition?

- Line 72/159: is not, not isn't; does not, not doesn't

- Line 75: shown, not as shown

- Line 96: Under Each or Under each

- Table 1: delete represents

- Line 125: previous gait cycle 

- Eq (11): too small to read

2. Many sections are only one sentence long, which does not make it a complete section at all, such as 2.2.2, consider rewrite.

3. One key concern is the ethical considerations of the data collected. In Line 87 the authors claim that "Subjects all are able-bodied and agree to participate in the data acquisition experiment." Please add further information on the IRB approval or related information.

4. The lack of description of the prothesis used. As claimed in paragraph 2, the paper is regarding data for control, then sufficient information must be provided to explain the mechatronic system and control method. But no further detail is provided at all, making it hard to make use of the method proposed and impossible to reproduce the results (no data is shared or open to the public). 

5. The main concern is the contributions claimed. The features used in this paper seem to appear in various publications in the past. Machine learning method helps, but for this particular research, the authors need to establish a baseline of comparison against something else. Also, why would these features produce a superior performance? What if someone changes to other features? Are these classifiers only tested with the data collected? Or are they further tested with the device used? Experiments, results, and discussions regarding the above questions are fundamental to establish the claimed contributions. 

6. The language or organization of this paper must be improved. 

- Paragraph 2 seems to start from nowhere, with no connection to Paragraph 1

- The reviews in Paragraph 3 is hard to read or shows no clear logic.

- If Figure 1 is a reproduction, then copyright issues must be explained and presented.

- Sentences such as line 240 show repetitively throughout this paper, making it more like technical documentation instead of a journal article. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This is a paper on control of leg prosthesis using five measured parameters processed by a neural network. As such it would seem of significance and in the end publishable. Some things need attention:

1. In Figure 2 it would help to define the meaning of the variables in the title. As there are five indicated here but the neural network, at Figure 14, takes 4M rather than 5M an explanation as to how one got lost would be in order. Also an indication of phi(.) might prove helpful. 

2. There are a very large number of abbreviations for which a table may prove helpful as one forgets near the end what was hidden in text at the start.

3. The reference: "Myoelectric control of robotic lower limb prostheses: a review of electromyography interfaces, control paradigms, challenges and future directions" by Aaron Fleming in the Journal of Neural Engineering would merit referencing. Especially since it has 152 references on this topic. 

4. In some of the figures it would help if the symbol of what is plotted were given. For example in Figure 8b, what are the symbols which go with the two axes (and also how is part b related to a)?

5. Comparing Figs. 6 and 8,  theta and its derivative are periodically related and since they enter in time in the equations, only one is actually needed. In any event it would help to understand Fig. 8 if theta were plotted in Fig. 6 along with the derivative (which is already there). 

7. In order for readers to repeat the results, the program listings used should be made available and the reader shown how to obtain them with a comment probably in the data availability statement (of which there should be also data made available). 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

No further comments