Using Mental Shadowing Tasks to Improve the Sound-Evoked Potential of EEG in the Design of an Auditory Brain–Computer Interface

Round 1

Reviewer 1 Report

Dear authors, despite the sample size, which we do not know if it is sufficient because you did not calculate the sample size power, it appears to be an interesting work.

 

Introduction

Lines 29-30: Replace the word ‘normal’ with ‘without dysfunction’.

Line 55: Replace the words ‘do something’ with ‘perform’.

It would be good to mention in the introduction or the discussion a work showing that performing an attention task in an intermittent noisy or silent room had different results. It is probably because humans are not conscious of all the noise surrounding them every second, but if we deprive them of it, it can be worse (Domingos et al. 2021).

 

Domingos, C., da Silva Caldeira, H., Miranda, M., Melício, F., Rosa, A. C., & Pereira, J. G. (2021). The Influence of Noise in the Neurofeedback Training Sessions in Student Athletes. International Journal of Environmental Research and Public Health, 18(24), 13223.

 

Methods

Line 159: Replace the word ‘are’ with ‘were’.

Lines 164-165: Rephrase ‘We will end the experimental procedure immediately and delete his data if a participant withdraws during the test’, with ‘The experimental procedure ended immediately, and the data were deleted if a participant withdraws during the test’.

Lines 200-201: Rephrase ‘We will then’ with ‘We used’.

Line 223: Replace ‘make’ with ‘made’.

 

Results

Please, do not highlight or bold your results. If they were significant, use the symbols (*).

You showed that you have a male or female voice, which is good, but it would be great to understand how different the girls in your study were compared to the boys.

 

 

Good luck!

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This study aimed to propose a BCI approach for Auditory Brain-Computer Interface with eeg signals. I have the following suggestions.

What is the novelty of this study although several approaches for BCI approach for Auditory Brain-Computer Interface with eeg signals have been proposed?

EEG is highly sensitive to the powerline, muscular, and cardiac artifacts. In EEG data preprocessing, authors need to mention how you handle AC power, ECG, and EMG artifacts in EEG signals. Same for EOG, EMG and others. Do the authors think that their proposed method is robust to such kinds of artifacts?

Authors should improve the conceptual figures of their ANN proposed frameworks with more details and model parametrization.

Which software/library do authors use to extract ERP?

CNN approach would be more convenient in this study.

Authors should introduce the EEG applications in ML-based disease, and mental workload prediction in broad scope, such as article, Explainable Artificial Intelligence Model for Stroke Prediction Using EEG Signal; in article, quantifying physiological biomarkers of a microwave brain stimulation device; in article, healthsos: real-time health monitoring system for stroke prognostics; in article, quantitative evaluation of task-induced neurological outcome after stroke; in article, driving-induced neurological biomarkers in an advanced driver-assistance system; and in article, quantitative evaluation of eeg-biomarkers for prediction of sleep stages.

The authors need to mention the model parameters or hyperparameters of ANN models. The performance of the model is dependent on the selection of the architecture and/or parameters.

Results should be a little bit concise. I recommend moving subject-specific results to the supplementary section. Overall results should be focused in main result section.

Authors should report more performance measures of ANN classifiers, such as sensitivity, specificity, precision, and negative predictive value from the confusion matrix.

Both cross-validated training and testing ROC curves should be reported.

I recommend using SHAP/LIME to explain the main contribution of ERP features in classification.

The discussion section needs to be included. Authors must make discussion on the advantages and drawbacks of their proposed method with other studies adding a table in the discussion section.

 

Clinical explanation of these findings needs to be described in support of reference.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Thank you for review responses. Several comments were not addressed carefully. 

1. The authors' noise removal/pre-processing response looks inappropriate. Eyeblink and EMG are different signals in terms of hardware circuitry. EEG electrodes can detect eye-blink, not EMG signals. It needs ICA algorithms or separate EOG electrodes to separate eye blink. I recommend using ICA to isolate and remove blink artifacts. 

2. Regarding ROC curves, I recommend combining all subjects in a feature matrix and performing ML classification of target vs non-target. Then ROC curves can be generated. 

3. Several other comments were not addressed. 

4. It is also recommended to report the ERP features of the target and non-target using an error-bar plot.

Author Response

Please see the attachment.

Author Response File: Author Response.docx