Review Reports

Round 1

Reviewer 1 Report

The article presents an interesting study on the applicability of the Attention Depression Detection Model Based on Multimodal Analysis. Nevertheless, before being accepted for publication, the article should be corrected according to the following guidelines:

• On what premise did the authors formulate the conclusion that: "In addition, the number of people diagnosed with depressive disorder in Korea 41 is steadily increasing"? The source is clearly missing here…
• The source for the following statement is also missing: "Table 1 is a descending order of data on people who received treatment for a depressive disorder by year from 2016 to 2020.”
• The authors did not explicitly indicate the aim of the article.
• The research gap must be better documented by referring to other works and showing what the authors' contribution really is in this regard.
• A new section, ‘discussion’, should be added to the article, where the results are related to the existing knowledge about the analyzed phenomenon.
• The article lacks information about contribution to existing knowledge (theory contribution), as well as practical implications.

Author Response

Thank you for your advice

Please see the attachment

 

   

 

 

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript does not link well with recent literature on AI for mental health, e.g., see sentic computing for patient centered applications. Also, latest trends in sequential fusion of facial dynamics for depression recognition are missing. Finally, check relevant works on mental disorder detection with attentive relation networks and Ji et al.’s recent review of suicidal ideation detection.

Author Response

Thank you for your advice

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper addresses the problem of handling and recognize emotional information by detecting depression. It is presented a depression detection model/system based on Multimodal Analysis. Fusion of text data and voice data is used in order to obtain a high-performance sensing model to improve the low accuracy of existing systems. The presented model and results might lead to considering that the approach is plausible. In fact, test results seem conclusive but the description of method and experimental results has flaws.

In the beginning the paper emphasizes the importance of this kind of models in the context of social media and related modern communication channels. Which is of highly interesting and deeply researched and explored today with all the useful, useless and controversial aspects of this kind of tech.

Sadly, in the end the paper abandons any hint about the applicability of the model and maybe possible scenarios where it could be applied, mentioning only IoT and AI speakers, but leaving no clue about how to use de results of the model.

Remarks:

- abbreviation in abstract incomplete - CNN-BiLSTM - Long Short-Term Memory

- it should be clarified what attention depression is and how it is used in this paper

- line 73 – unclear formulated statement

- 101 – unclear formulation

- figures have repeating parts

- 170-184 – text not related to the paper

- fig 3 – oversimplified with repetitions that are not explained or make no sense when not differentiated

- fig 4 – entitled ‘Problems with Fourier Transforms’ shows not the problem, it visually represents the transform. Very similar to

https://www.researchgate.net/figure/Diagram-of-the-time-domain-and-frequency-domain-of-the-signal_fig1_330203663

https://medium.com/analytics-vidhya/understanding-the-mel-spectrogram-fca2afa2ce53

- unclear how data are fused in the Fusion Model

- fig 7 – should contain a more in depth text description, not only by color. Two indistinguishable Dense blocks are also contributing to unclear repetitions

-258 – it should be mentioned that Ellie is not a robot or an AI but controlled by a hidden human

- section 5.2 – why are only sentences of the subjects pre-processed in this experiment? It should be argumentum why.

- formula (2) – all elements should be described

- the Experimental Results sections lacks detail analyses, it mainly contains several plots. A discussion is mandatory for such complex models. Mentioning that the ‘proposed system is significantly improved’ is not convincing.

- the paper uses ambiguously both terms: depression detection system and depression detection model

Therefore I consider that the paper “Design and Implementation of Attention Depression Detection Model Based on Multimodal Analysis” should be considered for resubmission only after major revision.

Author Response

Thank you for your advice

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear authors, congratulations on making all the suggested corrections. Compared to the previous version of the manuscript, I can see progress that has been achieved. Now the manuscript is more logical and coherent. Taking all this into account, I recommend the manuscript for publication.

Author Response

Thank you for your advice

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have addressed most of the concerns raised by the reviewers and their revisions have substantially improved the manuscript. However, there are still some minor issues to be addressed, namely:

1) presentation is better but there is still room for improvement

2) latest trends in sequential fusion of facial dynamics for depression recognition are still missing

3) relevant works on mental disorder detection with attentive relation networks should be added

4) recent efforts on emotion recognition based on reinforcement learning and domain knowledge are missing

Author Response

Thank you for your advice

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The revised version of the paper shows good improvement. The authors did address the remarks mentioned in the first review.

Therefore I consider that the paper “Design and Implementation of Attention Depression Detection Model Based on Multimodal Analysis” could be accepted in the current version.

Author Response

Thank you for your advice

Please see the attachment

Author Response File: Author Response.pdf