Photoplethysmographic Signal-Diffusive Dynamics as a Mental-Stress Physiological Indicator Using Convolutional Neural Networks
, J. Clemente 2,†, D. Fuentes-Jimenez 3,†, M. F. Cabrera-Umpiérrez 1,† and A. P. Gonzalez-Marcos 1,*,†
Round 1
Reviewer 1 Report
In this paper, a novel approach for analyzing the stress level based on the recorded PPg signals and its diffusive properties is presented. The manuscript is well-written and the authors tried to cover a wide range of studies provided in this field as the literature review. However, a list of concerns is listed below:
1- Problem definition: in recent real-world AI applications, where is the main potential of this study? I think these types of problems will gradually receive less attention compared to new applications of artificial intelligence. Also, I would like to ask the authors to declare the importance of automatic diagnosis of stress against the conventional methods used by physicians.
2- The preliminaries of the problem statement are too wordy. While being brief in stating the basics, Please focus on reviewing the studies regarding PPG.
3- Please clarify and list the main contributions of the study in the last parts of the introduction. To me, the reason why this manuscript could be considered a novel approach is the introduction of the p-q plane as the input of the CNN. Since CNN models can be considered as conventional techniques, how the state-of-the-art AI classifiers can make the method reliable?
4- According to Figure 2, please include the Results Section some p-q planes obtained from different states of the stress levels making the results legible. It is also can be necessary to add a corresponding discussion regarding the provided planes.
All the best.
Author Response
Our response is indicated by the blue letter, and any highlighted paragraphs are part of the new manuscript. The updated figures can only be found in the revised manuscript and are not included in the response to the reviewers.
The article draft underwent a thorough review, and we corrected the identified writing errors. Minor changes are not highlighted.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This work presented a deep learning model for detecting the presence of acute stress based on PPG signals. To improve this paper, the authors should improve this manuscript with the following comments.
1.) The introduction should be improved by compacting and providing a problem statement,
limitation of previous methods, the proposed method with reasons to solve the limitation.
The authors should summarize contributions of this work.
2.) Line 82 : The “PPG” is not defined before using.
3.) In section 2, please show some samples of PPG identifies as acute stress and others.
4.) In the section 2.2, what is the reason for using the 60-20-20 train-validation-test split validation
for evaluating the model?
5.) Fonts in Table 4 are too small. Please enlarge them.
6.) Do an ablation study.
7.) Give more insight why the proposed method performed well.
Author Response
Our response is indicated by the blue letter, and any highlighted paragraphs are part of the new manuscript. The updated figures can only be found in the revised manuscript and are not included in the response to the reviewers.
The article draft underwent a thorough review, and we corrected the identified writing errors. Minor changes are not highlighted.
Author Response File: Author Response.pdf
Reviewer 3 Report
I recommend that the authors consider other metabolic variables to demonstrate that physiological responses are associated and correlate with metabolic indicators. In addition, that they add to the discussion some proposals on the metabolic responses to different situations of acute stress.
Author Response
Our response is indicated by the blue letter, and any highlighted paragraphs are part of the new manuscript. The updated figures can only be found in the revised manuscript and are not included in the response to the reviewers.
The article draft underwent a thorough review, and we corrected the identified writing errors. Minor changes are not highlighted.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors have addressed all the previous comments. This manuscipt has been improved.
