All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Dear authors, we are pleased to verify that you meet the reviewer's valuable feedback to improve your research.
Thank you for considering PeerJ Computer Science and submitting your work.
Kind regards
PCoelho
[# PeerJ Staff Note - this decision was reviewed and approved by Mehmet Cunkas, a PeerJ Section Editor covering this Section #]
-no
-no
-no
-no
No more comments
No more comments
No more comments
No more comments
Dear authors,
You are advised to critically respond to the reviewer's comments point by point when preparing a new version of the manuscript and while preparing for the rebuttal letter.
Please address all comments/suggestions provided by reviewers, considering that these should be added to the new version of the manuscript.
In addition, I would ask that the authors edit the title to use person-first language. That is, change "autistic children" to "children with autism".
As a final note, a reviewer has suggested that you cite specific references. You are welcome to add it/them if you believe they are relevant. However, you are not required to include these citations, and if you do not include them, this will not influence my decision.
Kind regards,
PCoelho
No comments.
No comments.
No comments.
Thank you very much and I appreciate the fact that my opinions were taken into account.
I appreciate the fact that the authors took into account my suggestions and the article was completed, including the correct way to write the bibliographic sources.
I believe that the article can be published, but it should be better documented, to bring more arguments for each statement, at least in my opinion as a teacher and sociologist.
The paper contains statements that are not supported by the results of research, without reference to research in the field of education of children with autism spectrum disorders and the need for more realistic models to quickly and correctly diagnose this social category.
That is why I made the statement that there is a need to supplement it with information about the national educational context and the stage of national research on the school inclusion of students with autism spectrum disorders. As researchers in the field of education, I think you understand this argumentation, necessity and presentation of the state of research in your country.
The article has a relatively limited bibliography, it should have a more rigorous documentation (with more sources in the bibliography), with references to the results of studies published in different journals, such as, for example, Review Journal of Autism and Developmental Disorders, International Journal of Developmental Disabilities, etc…or articles: https://doi.org/10.7717/peerj-cs.1792; https://doi.org/10.3390/su13137056 ; https://doi.org/10.1080/20473869.2022.2070418.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors agree that they are relevant and useful.
I appreciate that the correct way of writing the bibliographic sources has also been corrected.
Good luck and congratulations for the work done.
See 4. Additional comments.
See 4. Additional comments.
See 4. Additional comments.
As I review this paper in its revised form, I find the study to be interesting. However, I have a major concern regarding the paper's proposal for using "Federated Learning," which requires multiple sources of network data. The primary content focuses predominantly on precision, recall, accuracy, and F-score, while aspects such as network processing, delays, and speed have not been addressed. Therefore, I strongly recommend either removing the mention of Federated Learning from the paper or providing extensive performance evaluations related to it, as I have outlined.
Dear authors,
You are advised to critically respond to all comments point by point when preparing an updated version of the manuscript and while preparing for the rebuttal letter. Please address all comments/suggestions provided by reviewers, considering that these should be added to the new version of the manuscript.
Kind regards,
PCoelho
-
-
-
This paper proposes a hybrid model that integrates a modified Convolutional Neural Network (CNN) for feature engineering, a Long Short-Term Memory (LSTM) network for predicting Autism Spectrum Disorder (ASD), and the DragonFly Optimization (DFO) algorithm to achieve better accuracy and performance.
However, there are several significant flaws in the study. Firstly, the use of Federated Learning (FL) is questionable because the dataset is shared in the .arff format, which is typically used directly in WEKA (an open-source machine learning software). This approach represents a traditional method and does not constitute true Federated Learning, as repeatedly claimed in the abstract and main text.
Furthermore, clarification is needed on how the .arff data file is utilized as input for the CNN in the feature selection stage. The data is tabular, while the authors mentioned in subsection 3.1 (Architecture) that it passes through a feature selection stage using CNN before being forwarded to the LSTM. This raises questions: Which CNN is being used? CNNs are primarily designed for image data, while the input here is tabular. Additionally, while LSTMs are suited for time series data, the authors are working with tabular data, which seems contradictory.
Moreover, the DFO optimization algorithm is said to adjust network parameters to minimize prediction errors and enhance accuracy. However, this process is not clearly explained. Overall, the methodology described contains several contradictions and major flaws.
There is also no source code provided to validate their model. It is essential for submissions to “AI Application” in PEERJCS to include source code. I do not understand why this screening phase was not conducted, especially since PEERJCS generally has a strict and complicated submission process that requires numerous edits and the sharing of source codes and datasets.
In conclusion, due to these substantial flaws and unsupported claims in the study, I lack confidence in their methodology and results. Therefore, I recommend rejecting this paper and leaving the final decision to the editor.
The article is topical, the subject being of interest to specialists, parents, teachers, or political decision-makers.
- In my opinion, the objective of the research should be clarified, and what are the research questions that were the basis of the practical approach? The objective should be specified from the beginning of the research, in the introduction and/or in the abstract.
The work is very interesting; it emphasizes the importance of federated learning (FL) for collecting ASD data and training models in the early detection of ASD and finding solutions.
In my opinion, in the theoretical part, arguments should be brought for the choice of the theme, the necessity of this research, in a national but also an international context. It should be clarified what the stage of research on this topic is in the educational national context, as well as the practical necessity of choosing this research topic.
In the research part, in my opinion, additional information should be clarified and brought related to:
- the research questions or research hypotheses, which were pursued in the empirical approach, were not clearly formulated;
- the studied population: - the population studied: how large were the data sets? Where were the data sets collected from, and how? Was the population of young children and schoolchildren participating in a screening program? How serious were the conditions of the children included in the research? This aspect is important for qualitative research: if children are involved in recovery therapies and do not have a severe diagnosis of the disorder, educational and social integration may be easier.
Research results: screening and prediction for ASD, the results should be analyzed and explained more from the perspective of practical implications for parents, teachers, management of educational institutions, or social policies in the field of education.
The accuracy of 99.23% obtained by ALROH for the dataset for children with ASD surpassed other models (CNN-LSTM-PSO, RF-XGB, XGB 2.0, SVM, RF, and L-R), which creates good premises for its use as an accurate and efficient method for detecting ASD. For the accuracy of the results, in my opinion, perhaps some correlations of the answers with data about the student's family (e.g., education level, economic level) or children (e.g., age and diagnosis, how long they have been in therapy) would validate the research instrument and provide greater objectivity.
The work is very interesting; it emphasizes the importance of federated learning (FL) for collecting ASD data and training models in the early detection of ASD and finding solutions.
It is honest for researchers to specify some limits, but I believe that providing future research perspectives would lead to clarification and understanding of the choice of research methods for this stage.
A major limit remains data collection and data accuracy, because we are talking about large databases, but also the confidentiality and security of the data flow regarding children and their health status. A limit remains on collaborative learning between different schools and the network of schools.
Good luck!
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.