Methods for Infectious Disease Risk Assessments in Megacities Using the Urban Resilience Theory

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In spite of the fact that the topic is interesting, well documented, I think the authors could improve their coherence in revealing their results. Consequently, I found certain paragraphs longer than needed. Furthermore, the discussion section could be extended.

Comments on the Quality of English Language

From my point of view, paraphrasing could  be used in order to avoid the repetition of a word more than three times in the same phrase. Moreover, I suggest the authors to use more connectors.

Author Response

Dear Reviewers:

 

Thank you very much for your time involved in reviewing the manuscript and your very encouraging comments on the merits. We also appreciate your clear and detailed feedback and hope that the explanation has fully addressed all of your concerns. In the remainder of this letter, we discuss each of your comments individually along with our corresponding responses.

Comments:

“In spite of the fact that the topic is interesting, well documented, I think the authors could improve their coherence in revealing their results. Consequently, I found certain paragraphs longer than needed. Furthermore, the discussion section could be extended.”

Response :

Thank you for the valuable feedback from the reviewer. We have split long paragraphs and discussed the applicability of the model, its usage, and limitations in the discussion section.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

In this article, the author employs Shanghai as a case study to illustrate the precision of geographic encoding technology in pinpointing granular patient distribution data as a pivotal factor. Additionally, the author combines points of interest (POI), population density (PD), and road network density (RD) as risk factors. Subsequently, through the utilization of geographic weighted regression, the author establishes a relational framework between these factors and the incidence of newly reported cases, thereby constructing a risk assessment model for evaluating infection risks across different regions. Notably, the author identifies that considering the interactions among these risk factors enhances the explanatory power of the spatial distribution of new cases, signifying synergistic effects. Particularly, the road network exhibits nonlinear reinforcement following its interaction with hotspots. These arguments hold significant relevance to the author's proposition of "administrative region-based risk spatial distribution." However, there are several issues that warrant resolution:

 

Major Concerns:

 

 

1. Selection of Shanghai as a Study Subject: The choice of Shanghai as the primary research subject may introduce bias. Factors such as population density, regional divisions, and healthcare infrastructure in Shanghai may not be universally applicable. It is essential to consider data from lower-density cities for experimentation. Furthermore, the keyword "data-driven" implies that the model derived from big data computations can be generalized to other cities. The study should elaborate on the model's generalization performance.

 

2. Suitability of Geographic Weighted Regression (GWR): The use of Geographic Weighted Regression (GWR) in this study may raise concerns. GWR typically requires a sufficiently large sample size to support the selected geographic area. Spatially uneven sample distribution can lead to estimation bias in spatial patterns. The article should address these considerations to establish the suitability of GWR for this experiment.

 

3. Introduction of "Urban Resilience" Concept: The introduction of the "urban resilience" concept should be substantiated with more concrete and reliable data comparisons before and after its implementation. Additionally, potential conflicts between the introduced method and baseline methods need clarification.

 

4. Consideration of Temporal Dynamics: The study mentions the LSTM model's sensitivity to temporal changes but does not provide an explanation at the temporal level. For assessing infectious disease risk, time is a crucial factor. The study should incorporate temporal information from the data to enhance the accuracy of infectious risk assessment.

 

5. Validation in Diverse Scenarios: The article can benefit from dividing Shanghai into various administrative regions and applying the same methodology to multiple regions. This would allow for the validation of the proposed approach across different scenarios and test the model's applicability in various contexts.

 

6. Lack of Citations and Comparative Analysis: Throughout the article, the author predominantly presents their own methods and results without sufficient referencing or discussion of related research. This diminishes the comprehensiveness and innovation of the study. For example, in the second paragraph of Introduction, PMID: 36939781 could be cited as the reference.

 

 

Minor Concerns:

1. Inconsistent Formula Numbering and Formatting: Ensure uniform formatting and numbering for equations, and maintain consistent font styles in formulas.

 

2. Review and Standardize References: Review all references for formatting consistency, correct formatting of volume numbers (e.g., "0x" or "x"), and ensure that references are properly numbered.

 

3. Figure Numbering and Labels: Correct any discrepancies in figure numbering, ensure consistent font styles for figure labels, and add distance legends or scale bars for maps.

 

4. Missing Table: Address the absence of Table in the manuscript.

 

5. Grammar and Punctuation: Pay attention to grammatical and punctuation issues, such as capitalization of "while" and redundant punctuation marks in the text.

 

6. Formatting Errors: Remove any unnecessary extra blank lines or formatting errors, especially after "3.result."

Comments on the Quality of English Language

Moderate editing of English language required.

Author Response

Dear Reviewers:

 

Thank you very much for your time involved in reviewing the manuscript and your very encouraging comments on the merits.

Comments:

“In this article, the author employs Shanghai as a case study to illustrate the precision of geographic encoding technology in pinpointing granular patient distribution data as a pivotal factor. Additionally, the author combines points of interest (POI), population density (PD), and road network density (RD) as risk factors. Subsequently, through the utilization of geographic weighted regression, the author establishes a relational framework between these factors and the incidence of newly reported cases, thereby constructing a risk assessment model for evaluating infection risks across different regions. Notably, the author identifies that considering the interactions among these risk factors enhances the explanatory power of the spatial distribution of new cases, signifying synergistic effects. Particularly, the road network exhibits nonlinear reinforcement following its interaction with hotspots. These arguments hold significant relevance to the author's proposition of "administrative region-based risk spatial distribution." However, there are several issues that warrant resolution”

 

We also appreciate your clear and detailed feedback and hope that the explanation has fully addressed all of your concerns. In the remainder of this letter, we discuss each of your comments individually along with our corresponding responses.

 

Comments1:

“Selection of Shanghai as a Study Subject: The choice of Shanghai as the primary research subject may introduce bias. Factors such as population density, regional divisions, and healthcare infrastructure in Shanghai may not be universally applicable. It is essential to consider data from lower-density cities for experimentation. Furthermore, the keyword "data-driven" implies that the model derived from big data computations can be generalized to other cities. The study should elaborate on the model's generalization performance.”

Response 1:

I would like to express my sincere gratitude for your valuable feedback. Shanghai, as one of China's major metropolitan areas, offers a high level of data accessibility, making it an ideal choice for our research. To assess the model's generalizability, we divided Shanghai into two distinct regions: a densely populated core area and other regions, subjecting each to separate validation. It is important to note that our experiments were solely conducted in Shanghai, thus our model's applicability was discussed in the title section and elaborated upon in the discussion section, primarily within the context of large urban settings.

Comments2:

“Suitability of Geographic Weighted Regression (GWR): The use of Geographic Weighted Regression (GWR) in this study may raise concerns. GWR typically requires a sufficiently large sample size to support the selected geographic area. Spatially uneven sample distribution can lead to estimation bias in spatial patterns. The article should address these considerations to establish the suitability of GWR for this experiment.”

Response 2:

Your insight regarding the application of the Geographically Weighted Regression (GWR) model at the city-level grid scale is greatly appreciated. Previous research employing GWR models for regression analysis in urban settings has demonstrated promising results. We ensured an adequate sample size and employed a uniform 1 km grid sampling approach in our study area, which likely mitigated the challenges posed by spatial non-stationarity and reduced the influence of outliers.

Comments3:

“Introduction of "Urban Resilience" Concept: The introduction of the "urban resilience" concept should be substantiated with more concrete and reliable data comparisons before and after its implementation. Additionally, potential conflicts between the introduced method and baseline methods need clarification.”

Response 3:

I am thankful for your comments regarding the theoretical basis of our research, which draws upon urban resilience theory. We selected indicators such as population density, Points of Interest (POI), and road networks to characterize a city's resilience in the face of a pandemic. The introduction of new cases served as a proxy for stress. While baseline models like LSTM or SEIR are theoretically capable of achieving high accuracy, particularly with a known basic reproduction number, our study focused on assessing variations in risk indices across different administrative regions within a city. Given the distinct focus and limited comparability between our approach and time-series models, we did not conduct direct comparisons with them.

Comments4:

“Consideration of Temporal Dynamics: The study mentions the LSTM model's sensitivity to temporal changes but does not provide an explanation at the temporal level. For assessing infectious disease risk, time is a crucial factor. The study should incorporate temporal information from the data to enhance the accuracy of infectious risk assessment.”

Response 4:

Time is indeed a critical variable, and our analysis was based on data collected at evenly spaced intervals. Our primary objective was to analyze variations in risk across different geographical areas, and as such, we did not compare risk models across different time periods within the same location. Additionally, our model evaluated the relative magnitudes of risk values for different regions within Shanghai over a 14-day forecast horizon, a short-term risk prediction task where the impact of time is relatively minor.

Comments5:

“Validation in Diverse Scenarios: The article can benefit from dividing Shanghai into various administrative regions and applying the same methodology to multiple regions. This would allow for the validation of the proposed approach across different scenarios and test the model's applicability in various contexts.”

Response 5:

In response to your feedback, we have supplemented the accuracy validation section of our results analysis. We divided Shanghai into core and non-core areas to assess the model's accuracy.

Comments6:

“Lack of Citations and Comparative Analysis: Throughout the article, the author predominantly presents their own methods and results without sufficient referencing or discussion of related research. This diminishes the comprehensiveness and innovation of the study. For example, in the second paragraph of Introduction, PMID: 36939781 could be cited as the reference.”

Response 6:

Lastly, I greatly appreciate your suggestion to compare our research with current studies and the citation reference (PMID: 36939781). We have incorporated this comparison in the discussion section and have referenced the mentioned article in the introduction。

 

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors A cordial greeting; congratulations to the team of researchers. It seems to me that the question addressed is of great value, for the application in public health, using novel methods from the statistical approach. The only question I would raise to improve the document is If it is possible to generate a table of other variables that could have been taken into account in the model with their specific weight and what is the reason to finally choosing these and not any other in the scenario of the urban resilience concept. I appreciate the opportunity of assessing the manuscript.  

Author Response

Dear Reviewers:

 

Thank you very much for your time involved in reviewing the manuscript and your very encouraging comments on the merits. We also appreciate your clear and detailed feedback and hope that the explanation has fully addressed all of your concerns. In the remainder of this letter, we discuss each of your comments individually along with our corresponding responses.

Comments:

“A cordial greeting; congratulations to the team of researchers. It seems to me that the question addressed is of great value, for the application in public health, using novel methods from the statistical approach. The only question I would raise to improve the document is If it is possible to generate a table of other variables that could have been taken into account in the model with their specific weight and what is the reason to finally choosing these and not any other in the scenario of the urban resilience concept. I appreciate the opportunity of assessing the manuscript.”

Response:

We appreciate the valuable feedback from the reviewer. We have added a table of model weights in the results section and discussed why these metrics were chosen over others in the discussion section.

 

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

Major comments:

1. Why were GLD, PD, POI, and RD selected as risk factors? Please explain your rationale.

2. Considering Crichton’s Risk Triangle (hazard, exposure, and vulnerability), there are other factors that directly contribute to the development of the outbreak. What about other potential risk factors? e.g., housing concentration, healthcare density, SES, age of population, etc. Please comment on why don’t you consider other factors.

3. Multicollinearity – it is clear to see the high correlation between those risk factors, hence the risk of multicollinearity. Please consider this in the analysis or at least in the discussion.

4. The scale of the variables were not unified. Normalization/scaling of variables may be needed.

5. The risk factors used in model training and prediction were essentially not changed in the short time period that was chosen in this study (weeks apart, I have to assume roads etc. were essentially the same). Hence, it is hard to convince the audience of the power of the model.

6. Do you consider the possibility of over-fitting? Or how is the generalizability of this risk assessment model? Independent test data should be used to verify the model instead of the same dataset. Otherwise, I don't see the potential applicability of this model. 

7. Please add captions for all figures.

8. Please edit figures using appropriate font and size and make sure the content is readable.

 

Minor:

1.     Line 105: please edit this sentence,  which is not correct in grammar. Please edit the language throughout.  

2.     Line 162: Figure 2 not figure 1. Add figure caption

3.     Line 167: I think you would like to refer to Figure 3 a

4.     Line 265: figure quality not good enough. Change the labeling to a larger font size

Comments on the Quality of English Language

Extensive editing of English language required

Author Response

Dear Reviewers:

 

Thank you very much for your time involved in reviewing the manuscript and your very encouraging comments on the merits.

We also appreciate your clear and detailed feedback and hope that the explanation has fully addressed all of your concerns. In the remainder of this letter, we discuss each of your comments individually along with our corresponding responses.

Comments1:

“Why were GLD, PD, POI, and RD selected as risk factors? Please explain your rationale.”

Response 1:

I would like to express my sincere gratitude for your valuable feedback. Our team has incorporated the reasons for selecting the data sources in the article's data section. The primary reason for choosing these data sources is that they draw upon related research indicating that pandemic risk is primarily influenced by these four factors. Additionally, these data sources are cost-effective and still yield a commendable level of accuracy in risk assessment.。

Comments2:

“Considering Crichton’s Risk Triangle (hazard, exposure, and vulnerability), there are other factors that directly contribute to the development of the outbreak. What about other potential risk factors? e.g., housing concentration, healthcare density, SES, age of population, etc. Please comment on why don’t you consider other factors.”

Response 2:

I also appreciate your input regarding the factors that were not included in our study. The main reasons for excluding these factors are as follows: firstly, these data sources often have poor availability and are typically challenging to obtain at high quality and spatial resolution. Secondly, previous research has shown that patient distribution, population density, road network density, and POI density are the primary drivers of pandemic risk. Hence, these four data sources were selected for this study. We have discussed the reasons for this choice in the discussion section of the paper.

Comments3:

“Multicollinearity – it is clear to see the high correlation between those risk factors, hence the risk of multicollinearity. Please consider this in the analysis or at least in the discussion.”

Response 3:

I want to thank you for your comment on the multicollinearity issue. After thorough examination by our team, the VIF factors between these four indicators are 6.7, 3.8, 4.4, and 2.7, all of which are below 10, indicating that there is no significant multicollinearity problem. We have also provided an explanation of the VIF results in the data section of the paper.

Comments4:

“The scale of the variables were not unified. Normalization/scaling of variables may be needed.”

Response 4:

Your suggestion regarding the standardization of input factors using the min-max method before geospatial detection and GWR regression has been duly considered. We have included an explanation of the standardization formula used in Equation 2 of the research methodology section.。

Comments5:

“The risk factors used in model training and prediction were essentially not changed in the short time period that was chosen in this study (weeks apart, I have to assume roads etc. were essentially the same). Hence, it is hard to convince the audience of the power of the model.”

Response 5:

I appreciate your insights into considering the temporal changes in patient distribution and assessing the risk in different regions over short periods. In the validation process, we have indeed used new patient distribution data to make predictions and compared them with real patient distribution.

Comments6:

“Do you consider the possibility of over-fitting? Or how is the generalizability of this risk assessment model? Independent test data should be used to verify the model instead of the same dataset. Otherwise, I don't see the potential applicability of this model..”

Response 6:

Regarding the GWR model's strong spatial heterogeneity in weightings and its effective fitting through local regression, we acknowledge your point. If this research model were to be applied in various regions, each region would need to utilize local epidemic data to establish models following the methodology outlined in this paper.。

 

 

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have no other questions.

Author Response

Thanks to the experts for their valuable comments on the revisions, the main revisions are to use the MDPI service to polish the English language of the article, and the proofs of the editing are in the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

These two concerns were not addressed in the first round of review

1. Please add captions for all figures.

2. Please edit figures using appropriate font and size and make sure the content is readable.

Comments on the Quality of English Language

English edits needed

Author Response

Thanks to the experts for their valuable comments on the English proficiency and charts of the article.

1 for the opinions on the charts, this revision will unify the annotation and font of the figures and fonts of the article.

2 for the English proficiency and readability opinions, this time the MDPI service has been used to polish the English article to increase readability，the polish certificate is shown in the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 4 Report

Comments and Suggestions for Authors

Please add captions for all figures.

And to the editors, please make sure the format is correct. 

Comments on the Quality of English Language

NA

Author Response

Dear Reviewers:

Thank you very much for your time involved in reviewing the manuscript.We also appreciate your clear and detailed feedback and hope that the explanation has fully addressed all of your concerns. In the remainder of this letter, we discuss each of your comments individually along with our corresponding responses.

Comments1:

“Please add captions for all figures.”

Response 1:

Thank you for the editor's feedback. I have added captions to all figures.

Comments2:

“And to the editors, please make sure the format is correct.”

Response 2:

Thank you for your valuable feedback. I have modified the format according to the template.