Assessment of Machine Learning Techniques in IoT-Based Architecture for the Monitoring and Prediction of COVID-19

Round 1

Reviewer 1 Report

Please see my comments below:

1. Looks like this paper cites data until August 2020 and every day new information and data are available regarding COVID19. For example, the authors mention in the introduction that "Sadly, there is no 26 reliable cure process or vaccination yet. The production of an efficient vaccination is
expected to take more than a year", which is no longer true. Hence, the paper requires extensive revisions in these aspects and more recent data/scenario regarding COVID19 need to be provided. 

2. Like I mentioned above, the hypothesis should be based on more recent data and literatures. The whole world is affected by COVID19 and everyday new research is getting published along this line. So, authors really need to cite some recent works from 2021. 

3. The figures are kind of discorganized and it's hard to follow. The figures/tables should be placed close to where they are first referred to in the text.

4. What is the novelty of this project? I could find similar work that are already published. One example is provided below:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7428786/ 

This raises concern about the novelty of this work. 

 

Author Response

Reviewer #1:

1. Looks like this paper cites data until August 2020 and every day new information and data are available regarding COVID19. For example, the authors mention in the introduction that "Sadly, there is no 26 reliable cure process or vaccination yet. The production of an efficient vaccination is expected to take more than a year", which is no longer true. Hence, the paper requires extensive revisions in these aspects and more recent data/scenario regarding COVID19 need to be provided.

• As suggested by Reviewer 1, the revised paper has been updated with recent information till July 2021 concerning the vaccination of COVID-19 for a better understanding of the reader.
• Moreover, the Literature Review Section of the revised manuscript has been completely updated to depict the state-of-the-art survey of research performed in the current domain.
• Furthermore, the recent scenario of COVID19 in terms of Delta and Delta+ variants has been provided in the revised manuscript for improving the quality of the paper.

 

2. Like I mentioned above, the hypothesis should be based on more recent data and literatures. The whole world is affected by COVID19 and everyday new research is getting published along this line. So, authors really need to cite some recent works from 2021.

• As advised by Reviewer 1, a state-of-the-art literature review has been performed in the revised manuscript for a better understanding of the reader.
• Specifically, the revised paper has been updated with the recent data of 2021 for improving the quality of the paper.
• Furthermore, the latest papers have been cited from 2020 and 2021 for an in-depth understanding of the reader.

 

3. The figures are kind of disorganized and it's hard to follow. The figures/tables should be placed close to where they are first referred to in the text.

• As advised, figures and tables are placed near the referred text in the revised paper for a better understanding of the reader.

 

 

4. What is the novelty of this project? I could find similar work that are already published. One example is provided below: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC 7428786/  This raises concern about the novelty of this work.

• As advised by Reviewer 1, the state-of-the-art research contribution has been depicted in Section 1.1 of the revised manuscript for a better understanding of the reader.
• Specifically, the IoT-Fog-Cloud framework is presented to predict the outspread of COVID-19 of geographical distribution.
• Moreover, the presented model incorporates several prediction techniques over the Fog-cloud platform for the detection and monitoring of the disease.
• Furthermore, the proposed model is dedicated to assessing the optimal prediction model for the early prediction of COVID-19 cases. It has been updated in the revised manuscript for a better understanding of the reader.
• Additionally, Table 2 of the revised manuscript has been updated to depict the novel contribution of the presented model in comparison to the state-of-the-art research work.

Author Response File: Author Response.pdf

Reviewer 2 Report

ENGLISH WRITING

English writing should be checked and corrected with detailed approach. Example 1: "it can provision" should be "it can provide", example 2: "3.1.3. Data Analytic" should be "Data Analytics"...Example 3: "this pre-processed dataset For our..." should have the word for with small f. Example 4: Receiver Operational Characteristic should be Characteristics, Example 5: A cross-Validation" should be "A cross-validation" with small v.

TITLE AND CONTENT

In the abstract, this paper announces results with simulation conduced, in aim to verify using algorithms in machine learning for specific use in COVID-19 symptoms processiong (for prediction use). The title announces IoT architecture for detection and tracking of COVID-19. The results of the paper show table and diagrams with comparison of prediction success for different machine learning algorithms. It is very important to harmonize title with the content of the paper, since the main results of the paper are related to machine learning algorithms, not IoT. If IoT is in title of this paper, then the whole content of this paper should be written differently, having focus on sensors and data acquisition, which is the core of IoT systems.

PAPER STRUCTURE

The paper does not have appropriate structure. It should have subtitles: Introduction, Theoretical Background, Related work, Proposed Approach, Research methodology, Empirical Results, Discussion, Conclusion, References.  

DETAILS

Figure 2. The photos of devices that are used in IoT, in embedded programming, but these devices belong to IoT level, not FOG level. Why this figure has been named Fog computing devices...? It is not necessary and it is not convenient to have photos of devices in scientific papers. Especially it is not necessary, since the title of this paper emphasizes Architecture of the system.

Figure 4. Conceptual framework does not have cloud service included. Secondly, it is very inprecise to have monitoring officials use FOG services directly (via access points, but it is not clear what are these access points made of)

Early prediction requires Big data processing and application of artificial intelligence algorithms to have valid conclusions as a support to decision making. The fundamental in decision making is to have a proper set of rules to be combined with data automatically and to process them in aim to get conclusions and decision making support. Figure 4. does not have CLOUD for data storage, BIG DATA processing, ARTIFICIAL INTELLIGENCE SOFTWARE to support decision making.

Section 3.1.3. states that data processing and machine learning algorithms has been utilized...but the section 3.1.3. belongs is 3.1. Presented framework...This section should present the approach (framework), not the implementation (utilization).

It is very important to have the proposed model presented separately from implementation details and from empirical research methodology. All these are mixed in the section 3. entitled "Proposed model". It is quite messy to have simulation and classification of data and prediction alltogether.

Figure 6. is entitled Confusion Matrix, but it has not been explained in nearby text. In fact, all figures in this paper have not been explained and should be explained in nearby text (better - closely before the figure occurence). 

3.3.1. Section Predictive model has explanation of different approaches in utilization of data prediction. These theoretical explanations should be moved into separate section entitled "Theoretical background", which should be located immediately after the introduction and before related work.

Research methodology should include more precise explanation of the hypothesis (not clearly stated), empirical research methods (it seems currently these methods are data prediction methods from machine learning), research sample. Currently, it is imprecise what is the sample - which symptoms and other data have been taken...It is too short to have this sample explanation: "15324 data instances of confirmed COVID-19 cases were acquired from the CORD- 19 repository. There are various forms of details about each case in the data. The current research focuses on signs, travel-history to vulnerable locations, and history of contact with COVID-19 patients."

Conclusions are not made according to the results. It is said in conclusion"The proposed method was used to create a machine-learning
based predictive model for illness, as well as to evaluate the therapeutic response, using possible COVID-19 case knowledge and clinical data of reported COVID-19 cases. The system also expresses these observations to healthcare doctors, who can then react rapidly to suspicious cases found by the predictive model by following up on any additional clinical examination necessary to validate the case. This helps the reported cases to be separated and proper health care provided." The paper presents only  results in evaluation of machine learning methods and  it does not provide results in therapeutic response, it does not provide a visualization for healthcare doctors to react in particular case of ill person....This paper simply does not have results as promised and concluded.

 

 

 

Author Response

1. ENGLISH WRITING

English writing should be checked and corrected with detailed approach. Example 1: "it can provision" should be "it can provide", example 2: "3.1.3. Data Analytic" should be "Data Analytics"...Example 3: "this pre-processed dataset For our..." should have the word for with small f. Example 4: Receiver Operational Characteristic should be Characteristics, Example 5: A cross-Validation" should be "A cross-validation" with small v.

• As suggested by Reviewer 2, the aforementioned mistakes have been corrected in the revised manuscript for improving the quality of the paper.
• Moreover, the revised paper has been read by me, and an English professor at the university for removing typo mistakes and grammatical errors.
• Furthermore, the online tool “Grammarly” has been used for eliminating errors and enhancing the English quality for improving the quality of the paper.

 

2. TITLE AND CONTENT

In the abstract, this paper announces results with simulation conduced, in aim to verify using algorithms in machine learning for specific use in COVID-19 symptoms processiong (for prediction use). The title announces IoT architecture for detection and tracking of COVID-19. The results of the paper show table and diagrams with comparison of prediction success for different machine learning algorithms. It is very important to harmonize title with the content of the paper, since the main results of the paper are related to machine learning algorithms, not IoT. If IoT is in title of this paper, then the whole content of this paper should be written differently, having focus on sensors and data acquisition, which is the core of IoT systems.

• As suggested by Reviewer 2, the title of the revised manuscript has been updated to “Assessment of Machine Learning Techniques for the Monitoring and Prediction of COVID-19” for harmonizing with the content of the paper.
• Moreover, the IoT has been used to depict the means to collect the data in real-time. It has been updated in the revised manuscript for a better understanding of the reader.

 

 

3. PAPER STRUCTURE

The paper does not have appropriate structure. It should have subtitles: Introduction, Theoretical Background, Related work, Proposed Approach, Research methodology, Empirical Results, Discussion, Conclusion, References.  

• As suggested by Reviewer 2, the revised paper has been re-organized as advised for improving the quality of the paper.

 

4. DETAILS

Figure 2. The photos of devices that are used in IoT, in embedded programming, but these devices belong to IoT level, not FOG level. Why this figure has been named Fog computing devices...? It is not necessary and it is not convenient to have photos of devices in scientific papers. Especially it is not necessary, since the title of this paper emphasizes Architecture of the system.

• As suggested, the photos have been removed in the revised manuscript for improving the quality of the paper.

 

Figure 4. Conceptual framework does not have cloud service included. Secondly, it is very inprecise to have monitoring officials use FOG services directly (via access points, but it is not clear what are these access points made of)

• As suggested, the details of Fog computing-based data assessment have been provided in the revised manuscript for a better understanding of the reader.

 

Early prediction requires Big data processing and application of artificial intelligence algorithms to have valid conclusions as a support to decision making. The fundamental in decision making is to have a proper set of rules to be combined with data automatically and to process them in aim to get conclusions and decision making support. Figure 4. does not have CLOUD for data storage, BIG DATA processing, ARTIFICIAL INTELLIGENCE SOFTWARE to support decision making.

 

Section 3.1.3. states that data processing and machine learning algorithms has been utilized...but the section 3.1.3. belongs is 3.1. Presented framework...This section should present the approach (framework), not the implementation (utilization).

It is very important to have the proposed model presented separately from implementation details and from empirical research methodology. All these are mixed in the section 3. entitled "Proposed model". It is quite messy to have simulation and classification of data and prediction alltogether.

Figure 6. is entitled Confusion Matrix, but it has not been explained in nearby text. All figures in this paper have not been explained and should be explained in nearby text (better - closely before the figure occurence). 

• As suggested, all the figures are placed along with referred text in the revised manuscript for a better understanding of the reader.

 

3.3.1. Section Predictive model has explanation of different approaches in utilization of data prediction. These theoretical explanations should be moved into separate section entitled "Theoretical background", which should be located immediately after the introduction and before related work.

• As suggested, the aforementioned change has been incorporated in the revised manuscript for improving the quality of the paper.

 

Research methodology should include more precise explanation of the hypothesis (not clearly stated), empirical research methods (it seems currently these methods are data prediction methods from machine learning), research sample. Currently, it is imprecise what is the sample - which symptoms and other data have been taken...It is too short to have this sample explanation: "15324 data instances of confirmed COVID-19 cases were acquired from the CORD- 19 repository. There are various forms of details about each case in the data. The current research focuses on signs, travel-history to vulnerable locations, and history of contact with COVID-19 patients."

• As advised by Reviewer 2, a precise explanation of the hypothesis has been provided in the revised manuscript for a better understanding of the reader.
• Moreover, the details regarding COVID-19 symptoms have been updated for improving the quality of the reader.

 

 

Conclusions are not made according to the results. It is said in conclusion"The proposed method was used to create a machine-learning
based predictive model for illness, as well as to evaluate the therapeutic response, using possible COVID-19 case knowledge and clinical data of reported COVID-19 cases. The system also expresses these observations to healthcare doctors, who can then react rapidly to suspicious cases found by the predictive model by following up on any additional clinical examination necessary to validate the case. This helps the reported cases to be separated and proper health care provided." The paper presents only results in evaluation of machine learning methods and it does not provide results in therapeutic response, it does not provide a visualization for healthcare doctors to react in particular case of ill person.... This paper simply does not have results as promised and concluded.

• As suggested, the conclusion of the paper has been updated according to the results derived in the paper.
• Moreover, the details about visualization have been provided in the revised paper for a better understanding of the reader.

Reviewer 3 Report

This work analyzes 8 machine learning techniques, namely Neural 11 Network, Decision Table, Support Vector Machine (SVM), Naive Bayes, OneR, KNearest Neighbor, 12 (K-NN), Decision Stump, and ZeroR, to detect corona-virus cases from time-sensitive information. It is an interesting work and well written. All methods are applied are machine learning, however, I would suggest to apply deep learning models too (atleast one).  for example: Depth-Wise Dense Neural Network for Automatic COVID19 Infection Detection and Diagnosis

Understanding Information Spreading Mechanisms During COVID-19 Pandemic by Analyzing the Impact of Tweet Text and User Features for Retweet Prediction, 

Improving coronavirus (COVID-19) diagnosis using deep transfer learning

Author Response

This work analyzes 8 machine learning techniques, namely Neural Network, Decision Table, Support Vector Machine (SVM), Naive Bayes, OneR, KNearest Neighbor, 12 (K-NN), Decision Stump, and ZeroR, to detect corona-virus cases from time-sensitive information. It is an interesting work and well written. All methods are applied are machine learning, however, I would suggest to apply deep learning models too (atleast one).  for example: Depth-Wise Dense Neural Network for Automatic COVID19 Infection Detection and Diagnosis

Understanding Information Spreading Mechanisms During COVID-19 Pandemic by Analyzing the Impact of Tweet Text and User Features for Retweet Prediction, improving coronavirus (COVID-19) diagnosis using deep transfer learning

• The author would like to thank Reviewer 3 for the positive comments.
• As suggested by Reviewer 3, an experimental section in the revised manuscript has been updated with the deep learning method of Dense Neural Network and Long short-term memory technique for improving the quality of the paper.
• Moreover, the aforementioned papers have been appropriately referred in the revised manuscript for a better understanding of the reader.

Reviewer 4 Report

The paper proposes a detection and tracking method of COVID-19 based on

IoT platform using machine learning (ML) techniques. As an application oriented

paper, this paper is lack of originality. All ML algorithms  are existing, as well as

the procedures such as symptoms  data collection, data analysis, etc are the

common steps in building predictive analytics model. This makes the paper’s

contribution is poor and not significant enough for getting published in this journal.

Author Response

The paper proposes a detection and tracking method of COVID-19 based on IoT platform using machine learning (ML) techniques. As an application-oriented paper, this paper is lack of originality. All ML algorithms are existing, as well as the procedures such as symptoms data collection, data analysis, etc. are the common steps in building predictive analytics model. This makes the paper’s contribution is poor and not significant enough for getting published in this journal.

• As advised by Reviewer 4, the revised paper has been updated with the deep learning techniques of Dense Neural Network, and Long short-term memory technique for improving the quality of the paper.
• Moreover, the revised paper has been updated with the fog computing environment for presenting the novel aspects of the proposed model.
• Furthermore, the revised manuscript of the presented model has been updated with the novel aspects of fog computing-based detection of the spread of COVID-19.
• Additionally, Subsection 1.2 has been made to depict the novel concepts of the presented model for a better understanding of the reader.

Round 2

Reviewer 1 Report

Thanks for the revisions. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

TITLE

Large portion of text still is related to complex IoT based system for data collection and processing, before the machine learning subsystem gets the data and enables prediction. Therefore, authors should consider having IoT-based architecture included in title, together with Machine Learning Techniques asessment...

INTRODUCTION

Introduction is too short, only one paragraph. There should be at least one more paragraph explaining the purpose and structure of the paper.

RELATED WORK

Related work contains both previous research results in applying IoT and machine learning in healthcare, with special focus on COVID-19. It is adviced to have related work presented with 2 subsections - IoT in healthcare, Machine learning in Healthcare.

PROPOSED APPROACH

Section 1.1.1. entitled PRedictive model is provided under Theoretical background and it contains some elements of the proposed approach, too early. It should be included within the Proposed approach section.

RESEARCH METHODOLOGY

Section 3.1. entitled Research methodology does not have hypotheses, research questions, but it does provide steps of the proposed approach combined with steps of empirical research. Research sample is explained within 3.2.1. Data Instances subsection.

CONCLUSION

Conclusion is not precise according to results. It states that "The system also provides data to healthcare doctors via a fog computing platform, who can then react rapidly to suspicious cases found by the predictive model by following up on any additional clinical examination necessary to validate the case." This is not obvious from the paper, since there is no software presented in this paper, that could illustrate the utilization of the proposed system. The whole system (based on IoT, FOG and even GIS) is hypothetic, but only COVID-19 dataset and machine learning techniques are true results of this paper.

REFERENCES

There is no Title "REFERENCES", but they start right after conclusion.

Author Response

Please see the attachment

Author Response File: Author Response.pdf