Physiological Signal Analysis and Stress Classification from VR Simulations Using Decision Tree Methods
Round 1
Reviewer 1 Report
The authors present a comprehensive physiological analysis of stress from an experiment involving a VR video game Bubble Bloom to manage stress levels. The idea is interesting, but the paper needs a careful writing review as many minor errors were found and will be listed in sequence. Please note the following comments are intended to improve paper quality and readers' understanding.
First page is blank. Please correct this.
The numbers of the sections are wrong. Please update them considering Introduction as section number 1.
The number of users that participated on the experiments is not clear. maybe a diagram would help understand what happened with the initial 14 users and why they were not able to participate on all phases.
Please provide more details regarding the different phases of the experiments as this is not clear in the text. What was the duration of each phase, for instance? What was the protocol used? How were the participants selected?
Please make more explicit what is the main contribution of the paper and how it advances the state of the art in stress detection from VR video games.
More general comments and minor errors are listed as follows.
"Personalized CART" -> "The personalized CART"
"XGBoost and DT was able" -> "XGBoost and DT were able"
"72.22% using" -> "72.22% accuracy using"
"0. Introduction" -> "1. Introduction"
"there in an increase" -> "there is an increase"
"for its ability" -> "for their ability"
"freqeuncy" -> "frequency"
"R-R interval that" -> "R-R intervals that"
"Kubios HRV but" -> "Kubios HRV, but"
"the model fit" - > "the model fits"
"signal processing an ML" -> "signal processing and ML"
"often lead to an" -> "often leads to an"
"where as a ratio above" -> "where a ratio above"
"a persons" -> "a person's"
"25, which were" -> "25, were"
"the actual actual output " -> "the actual output "
"Personalized CART model" -> "Personalized CART model:"
"in alot of" -> "in a lot of"
"index iteratively goes" -> "indexes iteratively go"
"calculates" -> "calculate"
"Decision Trees:" -> please correct identation
"Pipeline:" -> please correct identation
"Grid Search:" -> please correct identation
" maybe conducted" -> " may be conducted"
"Fig. 4demonstrates" -> "Fig. 4 demonstrates"
"Researcher’s often utilized" -> "Researchers often utilize"
" the baseline base to" -> " the baseline to"
"CART model was" -> "The CART model was"
"results from Table 3 illustrates" -> "results from Table 3 illustrate"
"respectively." -> ", respectively."
"it was. As certain" -> "it was, as certain"
" for a 2D" -> " for 2D"
"order effectively" -> "order to effectively"
"Fig.7" -> "Fig. 7"
"demonstrates a higher" -> "demonstrate a higher"
"was achieved" -> "were achieved"
"feature.We" -> "feature. We"
"using k-nearest neighbors" -> "using the k-nearest neighbors"
"but its an" -> "but it's an"
"with 87.75% using" -> "with a 87.75% of accuracy using"
"in the future. . " -> "in the future."
"Table 2 demonstrate" -> "Table 2 demonstrates"
Author Response
1) First page is blank. Please correct this.
The numbers of the sections are wrong. Please update them considering Introduction as section number 1.
Ans: We would like to thank the reviewer for taking the time to review and improve the content of our manuscript. We have fixed this issue, as the first page is no longer blank. It contains the abstract. The section number has also been fixed, with the introduction changing from 0 to 1.
2) The number of users that participated on the experiments is not clear. maybe a diagram would help understand what happened with the initial 14 users and why they were not able to participate on all phases.
Ans: Thank you noticing this. To resolve this, we have added a diagram, table and indicated why certain subjects were not able to participate, which is highlighted from line 154 - 190 in the manuscript.
3) Please provide more details regarding the different phases of the experiments as this is not clear in the text. What was the duration of each phase, for instance? What was the protocol used? How were the participants selected?
Ans: We would like to thank the reviewer for bringing this to our attention. We have addressed this issue and included a diagram to indicate the difference phases of the experiments.
4) Please make more explicit what is the main contribution of the paper and how it advances the state of the art in stress detection from VR video games.
Ans: We have addressed this as per the reviewers suggestion by including a contribution section to illustrate the significance of our work. This is highlighted from line 126 - 138.
5) More general comments and minor errors are listed as follows.
Ans: We have made all the required changes to fix the grammatical errors noticed by the reviewer and us. All the changes made were highlighted in yellow.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors have presented a method for stress classification which is based on ECG, GSR and RESP signals from 15 subjects. The VR Bubble bloom fish game has been applied for mitigating stress. I have few issues that they need to be considered before the paper can be accepted for publication.
-The authors have mentioned that K-means clustering is novel but I did not see any novelty in the proposed k-means algorithm which is just a traditional approach. In addition, I don't see any logic behind using it. What is the impact of using this part?
-Figure 3 has been extracted and inserted directly from Matlab I think, which is not proper for a scientific paper.
-From what I understand from the paper, the labeling has been done based on the features which is not true, I think. In this way it should be possible to classify the stress based on the same features which have been used for the labeling with high accuracy. The real labeling process should be done based on the task at different step of inducing the stress.
The writing can be edited by authors.
Author Response
Point-to-Point responses to the Reviewer
1) The authors have mentioned that K-means clustering is novel but I did not see any novelty in the proposed k-means algorithm which is just a traditional approach. In addition, I don't see any logic behind using it. What is the impact of using this part?
Ans: We would like to thank the reviewer for taking the time to review and improve the content of our manuscript. We have addressed this on line 240-245.
2) Figure 3 has been extracted and inserted directly from Matlab I think, which is not proper for a scientific paper.
Ans: Thank you for noticing this. To resolve this, we have removed the diagram and added the results obtained from the experiment in text. The description is highlighted on line 390-395.
3) From what I understand from the paper, the labeling has been done based on the features which is not true, I think. In this way it should be possible to classify the stress based on the same features which have been used for the labeling with high accuracy. The real labeling process should be done based on the task at different step of inducing the stress.
Ans: We would like to thank the reviewer for bringing this to our attention. We have addressed this issue and included the phase where we label the data based on the task rather than the feature. This is highlighted on line 341-347.
Author Response File: 
Author Response.docx
Round 2
Reviewer 1 Report
Dear authors, thank you for considering the comments. I'm satisfied with the modifications performed. Please take into consideration the following errors before producing the final version of the paper. Congratulations!
"Origianlly" -> "Originally"
"70% ," -> "70%,"
"outliers.[36]" -> "outliers[36]."
Reviewer 2 Report
The authors have replied to all issues raised in the first round of revision so I am happy to accept it for publication now.
