Wearable Sensor–Based Telerehabilitation Versus Conventional Physiotherapy in Knee OA: Insights from the KneE-PAD Pilot Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1.    The abstract does not clearly indicate the sample size and group allocation.
2.    In the abstract’s results section, the focus is only on the improvement of sEMG indicators, but the main clinical outcome, WOMAC, lacks specific values or the extent of improvement. These should be supplemented.
3.  The descriptions of rehabilitation interventions for the control and intervention groups mention that they are “based on guidelines,” but lack details about specific exercises, sets, and intensity.
4.    The results section provides insufficient description of the psychological assessment outcomes. Only the significance of HADS-D is reported, but detailed comparisons of TSK and ASES are missing and should be fully presented.
5.    The discussion focuses mainly on feasibility but lacks in-depth explanation of the negative findings.

Author Response

We sincerely thank the reviewer for the insightful and constructive comments, which helped us to improve the scientific clarity and completeness of the manuscript. Below, we provide a detailed point-by-point response. All modifications have been incorporated in the revised version and highlighted accordingly.

Comment 1

“The abstract does not clearly indicate the sample size and group allocation.”

Response:

We appreciate this comment. The abstract has been revised to specify both the total sample size and the group allocation.

Comment 2

“In the abstract’s results section, the focus is only on the improvement of sEMG indicators, but the main clinical outcome, WOMAC, lacks specific values or the extent of improvement. These should be supplemented.”

Response:

We agree and have added specific WOMAC outcome values in the Results section of the abstract

.Comment 3

“The descriptions of rehabilitation interventions for the control and intervention groups mention that they are ‘based on guidelines,’ but lack details about specific exercises, sets, and intensity.”

Response:

This valuable comment has been addressed by providing a more detailed description of the exercise protocols, including sets, repetitions, and intensity parameters. Moreover, for full transparency, all exercises performed by participants are presented in photographic form in Appendix of the revised manuscript.

Comment 4

“The results section provides insufficient description of the psychological assessment outcomes. Only the significance of HADS-D is reported, but detailed comparisons of TSK and ASES are missing and should be fully presented.”

Response:

We have expanded the Results section to include detailed pre- and post-intervention values, p-values, and effect sizes for TSK, HADS, and ASES.

Comment 5

“The discussion focuses mainly on feasibility but lacks in-depth explanation of the negative findings.”

Response:

We have expanded the Discussion section to include a more detailed interpretation of the limited psychological improvements and the absence of significant between-group differences.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a well-written and methodologically sound pilot study with promising results on wearable sensor–based telerehabilitation; however, the small sample size, limited AI integration, and overinterpretation of findings reduce its current scientific strength. The study has potential but requires clarification of methodology, stronger justification of claims, and improved data interpretation before acceptance.

1. The sample size (n = 20) is very limited for drawing meaningful conclusions; while acceptable for a pilot study, the authors should explicitly discuss the potential risk of type II error and limited generalizability.

2. The AI-driven postural error detection feature of the KneE-PAD platform was disabled during the trial. This weakens the technological novelty; the rationale for deactivation and its expected impact on outcomes should be elaborated.

3. Although the study claims single-blind design, participant blinding is impossible in behavioral interventions. The manuscript should clarify how assessor blinding was ensured and whether participants’ expectations could have biased self-reported measures.

4. The control and intervention groups followed the “same” exercise protocol, but supervision intensity and interaction differ substantially; this confounder should be addressed as it may influence adherence and psychological outcomes.

5. Several outcome measures show statistical improvement without significant between-group differences; the authors should interpret this cautiously and avoid overstating comparative effectiveness.

6. The inclusion of multiple parametric and non-parametric tests with small sample size increases risk of false positives; correction for multiple comparisons (e.g., Bonferroni) or discussion of type I error inflation is recommended.

7. Figures and tables present extensive numeric data, but visualization of key findings (e.g., effect sizes, pre-post changes) could improve clarity and accessibility for readers.

8. While the authors discuss clinical meaningfulness (MCID) appropriately, these thresholds are population- and context-specific; citing the exact validation studies or sensitivity analyses would strengthen credibility.

9. The discussion section interprets sEMG improvements as evidence of neural adaptation, but this inference requires caution, as sEMG amplitude can be influenced by electrode placement or signal normalization methods that are not fully described.

10. The manuscript lacks a clear usability or satisfaction metric (e.g., System Usability Scale, MAUQ); incorporating validated user experience tools would substantiate claims of high feasibility and acceptance.

Author Response

We sincerely thank the reviewer for the constructive and insightful comments that significantly helped improve the quality and clarity of our manuscript. Below, we provide a detailed, point-by-point response.

Comment 1 — Small sample size and type II error

The sample size (n = 20) is very limited for drawing meaningful conclusions; while acceptable for a pilot study, the authors should explicitly discuss the potential risk of type II error and limited generalizability.

Response: We thank the reviewer for this important observation. We acknowledge that the small sample size limits statistical power and generalizability. As this was a pilot trial, our primary aim was to assess feasibility and derive effect sizes for future larger studies.

Comment 2 — Disabled AI module

The AI-driven postural error detection feature of the KneE-PAD platform was disabled during the trial. This weakens the technological novelty; the rationale for deactivation and its expected impact on outcomes should be elaborated.

Response:We appreciate this comment and have clarified the rationale in Section 2.4 (Interventions). The AI module was intentionally disabled to validate the platform’s baseline feasibility and data stability without automated classification. This allowed a direct comparison between visual feedback and conventional physiotherapy without algorithmic influence.

Comment 3 — Single-blind design clarification

Although the study claims single-blind design, participant blinding is impossible in behavioral interventions. The manuscript should clarify how assessor blinding was ensured and whether participants’ expectations could have biased self-reported measures.

Response:We agree and have expanded Section 2.3 (Randomization and Blinding). Outcome assessors were fully blinded to group allocation, and participants were instructed not to disclose their treatment modality during evaluations.

Comment 4 — Confounding due to supervision intensity

The control and intervention groups followed the “same” exercise protocol, but supervision intensity and interaction differ substantially; this confounder should be addressed as it may influence adherence and psychological outcomes.

Response:We thank the reviewer for this insightful point. We now acknowledge in the Discussion that differences in supervision intensity may have influenced motivation, adherence, or psychological responses.

Comment 5 — Overinterpretation of non-significant between-group results

Several outcome measures show statistical improvement without significant between-group differences; the authors should interpret this cautiously and avoid overstating comparative effectiveness.

Response:We concur and have revised the Discussion to emphasize that improvements were comparable rather than superior across groups.

Comment 6 — Multiple comparisons / Type I error

The inclusion of multiple parametric and non-parametric tests with small sample size increases risk of false positives; correction for multiple comparisons (e.g., Bonferroni) or discussion of type I error inflation is recommended.

Response:We thank the reviewer and have added a note in Section 2.7 (Statistical Analysis) acknowledging this issue.

Comment 7 — Visualization of findings

Figures and tables present extensive numeric data, but visualization of key findings (e.g., effect sizes, pre–post changes) could improve clarity and accessibility for readers.

Response:

We thank the reviewer for this valuable suggestion. In response, two representative visualizations (Figures 26 and 27) have been added to the revised manuscript, illustrating the pre–post changes in functional (TUG, quadriceps strength) and neuromuscular (sEMG RMS, peak-to-peak) outcomes for both study groups. These figures complement the quantitative results in Tables 2 and 3 and enhance interpretability of the main findings. Additional figures were intentionally limited to maintain conciseness and focus on the key outcome domains most relevant to the study’s objectives.

Comment 8 — MCID contextualization

While the authors discuss clinical meaningfulness (MCID) appropriately, these thresholds are population- and context-specific; citing the exact validation studies or sensitivity analyses would strengthen credibility.

Response:We have now cited the corresponding validation sources for each instrument (WOMAC, TUG, ASES, HADS, TSK) and clarified that thresholds are based on Greek or comparable OA populations

Comment 9 — Neural adaptation and sEMG

The discussion interprets sEMG improvements as evidence of neural adaptation, but this inference requires caution, as sEMG amplitude can be influenced by electrode placement or signal normalization methods that are not fully described.

Response: We agree and have revised the Discussion to reflect this caution.

Comment 10 — Lack of usability metric

The manuscript lacks a clear usability or satisfaction metric (e.g., System Usability Scale, MAUQ); incorporating validated user experience tools would substantiate claims of high feasibility and acceptance.

Response:We appreciate this valuable feedback. We have acknowledged this as a limitation in Section 5 and proposed inclusion of validated usability tools (SUS, MAUQ, UTAUT2) in future research.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

In response to the previous review suggestions, the authors made good revisions.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed my previous major comments and concerns, and the manuscript has been revised accordingly. Therefore, I recommend the manuscript for acceptance and publication in its current form