A Time-Scalable Posture Detection Algorithm for Paraplegic Patient Rehabilitation Using Exoskeleton-Type Wearable Robots

Round 1

Reviewer 1 Report

Authors proposed an algorithm that can optimise rehabilitation outcomes by comparing posture data generated during the rehabilitation of a paraplegic patient wearing a body-tracking sensor with reference posture data. They concluded that with the proposed time scalable posture detection algorithm (TSPDA), the analysis target is not limited to the walking posture in posture analyses; the algorithm can universally accommodate various input postures using a full-body motion capture device.
The topic is very interesting and the paper is well structured and writed.

Author Response

We've edited the manuscript due to some feedback of other reviewers.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper addresses a problem of interest as related to the development of motion assistance devices based on exoskeletons. Attention to the related issues is limited to aspects, although interesting, but dealt with in a rather superficial way and from a descriptive point. Scaling up requirements and solutions is undoubtedly a promising idea to facilitate functional design and functionality checks before tackling the construction and testing of prototypes for demanding campaigns in healthcare fields. The work remains extremely descriptive and does not give any quantitative evaluation of reference parameters as well as the analysis of the state of the art is limited not only in the number of publications in the bibliography, but also in the systems and approaches that are used to develop systems of motion assistance, not only with exoskeletons, for the medical health field in which the work is focused on.

Author Response

A paper entitled “A Human-Robot Interface System for WalkON Suit: A Powered Exoskeleton for Complete Paraplegics” showed that exoskeleton-based wearable robot technology was evolving rapidly, but the learning curve remained steep. Our algorithm flattens the curve. We now make this point in Section 2

 

The “accuracies” are the differences between runs that did not feature slice counting and those featuring various slice counts. The graph of the optimal case (no slice count) is most similar to the reference graph; our quantitative evaluation is thus valid. We have deleted extraneous (confusing) data. Some early data were inaccurate; we have made corrections.

 

We have added recent sensor citations to Section 2; few works have explored the methodological aspects of inertial sensor measurement and processing.

 

As our TSPDA is based on operative principles only, we have added text (to the last section) that emphasises the social potential of TSPDA.

Author Response File: Author Response.pdf

Reviewer 3 Report

The work of Ho-won Lee et. al. presents many major issues. 

1) the introduction (sections 1 and 2) is poorly referenced, in particular, there is a lack in reporting the methodological aspects related to the inertial sensor measurements and processing techniques. I suggest the authors to review the following manuscripts

- "Magnetometer-Free Sensor Fusion Applied to Pedestrian Tracking: A Feasibility Study", IEEE ISCT 2019.
- "Estimation of human center of mass position through the inertial sensors-based methods in postural tasks: An accuracy evaluation", Sensors 2021.
-"Enhancing Athlete Tracking Using Data Fusion in Wearable Technologies", IEEE Access 2021.

2) It is not clearly stated the contribution of this work with respect to the literature and any comparison have been reported. 

3) The TSPDA architecture seems to be a preprocessing tool for multi- inertial sensors recordings, however the pseudo-codes do not help in the understanding of the methodological advancements proposed.

4)Figure reported in the results seem to be screenshot of the acquisition protocol. Moreover they are in a very poor resolution. I suggest the authors to better visualize the time series through MATLAB or Python.

5) It is not clear what the accuracy indicated in the tables of the result section. Indeed, authors stated aspect regard "significance" without employing any statistical test.

6)Consequently, the discussion part reported after the Results is poorly supported and difficult to be understood.

Author Response

Moderate English changes required

=> The English in this document has been checked by at least two professional editors, both native speakers of English. For a certificate, please see:
http://www.textcheck.com/certificate/gG0D8m

 

The introduction (sections 1 and 2) is poorly referenced, in particular, there is a lack in reporting the methodological aspects related to the inertial sensor measurements and processing techniques. I suggest the authors to review the following manuscripts

=> In terms of the IMU sensor literature, our original submission indeed lacked adequate information. We have corrected this deficiency. Thank you.

 

It is not clearly stated the contribution of this work with respect to the literature and any comparison have been reported. 

=> Unlike what was reported in previous studies, the TSPDA operates over a wide range of postures and the parameters can be tuned. We now emphasise these advantages in the last section of the paper.

 

The TSPDA architecture seems to be a preprocessing tool for multi- inertial sensors recordings, however the pseudo-codes do not help in the understanding of the methodological advancements proposed.

=> A pseudo-code is a modular operating principle. Section 3 clarifies that the combined modular functions (not the pseudo-codes) are described in the Results. If our “pseudo-code” approach renders understanding difficult, we will consider deleting all references to pseudo-codes.

 

Figure reported in the results seem to be screenshot of the acquisition protocol. Moreover they are in a very poor resolution. I suggest the authors to better visualize the time series through MATLAB or Python.

=> The screenshots were obtained from the Lua and SDL libraries. Indeed, resolution and readability were poor. We have dealt with these problems; line thickness was increased by modifying the code.

 

It is not clear what the accuracy indicated in the tables of the result section. Indeed, authors stated aspect regard "significance" without employing any statistical test.

=> The “accuracies” are the differences between runs that did not feature slice counting and those featuring various slice counts. The graph of the optimal case (no slice count) is most similar to the reference graph; our quantitative evaluation is thus valid. We have deleted extraneous (confusing) data. Some early data were inaccurate; we have made corrections.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors responded to all my concerns, despite the pseud-code remains still difficult to be interpreated.

The paper can be accepted for publication.