Revealing the Impact of Urban Form on COVID-19 Based on Machine Learning: Taking Macau as an Example
Round 1
Reviewer 1 Report
1.Please add a definition of urban form in the manuscript.
2.The spread of the new coronavirus and the transformation of the urban morphology are likely to be influenced by the human flow. Please add to this point.
3.Please also add how the morphology of the city will change after the spread of the new coronavirus is under control.
Author Response
Dear Reviewer,
We have revised our manuscript according to all of your comments. Edited text in the attached revised manuscript is visible as tracked changes under the All Markup mode of Microsoft Word.
(1)Comments: Please add a definition of urban form in the manuscript.
Response: We add the definition of urban form and the elements of urban form considered in this paper at the beginning of Section 2（line 90-97）. At the same time, due to the reference to the definition concept, we have also adjusted the order of references.
(2)Comments: The spread of the new coronavirus and the transformation of the urban morphology are likely to be influenced by the human flow. Please add to this point.
Response: Frankly, "The spread of the new coronavirus and the transformation of the urban morphology are likely to be influenced by the human flow. Please add to this point." is another research topic. In other words, the relationship between COVID-19 and urban form (material space elements) has not been revealed by further empirical research, at least in the field of urban planning or design. Therefore, our research mainly focuses on this angle.
(3)Comments: Please also add how the morphology of the city will change after the spread of the new coronavirus is under control.
Response: We are very grateful to the reviewer for this suggestion. According to the analysis of the elements of urban form (material space) in the current study, after the epidemic is under control, a greater degree of transformation may be green space. That is to say, the area of green space will increase with the process of urban planning in the future. This is also because the urban green space has a certain function of disaster prevention and hedging. We have pointed out this consideration in the second point of "4. Discussion : Pandemic and Sustainable Living".
Reviewer 2 Report
The paper makes a useful contribution, indicating that most machine-based learning applications in relation to the pandemic have dealt with epidemiological prediction rather than the impact of urban environmental factors on the pandemic. The means of sampling and the deployment of machine learning is interesting and innovative. In this sense, it is a worthy paper for possible future publication.
However, I do have a major query in relation to the paper. The Covid-19 population seems to be calculated on an absolute rather than per capita basis. If this is indeed so, the value of the paper is seriously undermined.
The conclusions reached in lines 256 – 261, for example, are self-evident in a way. The more people in an area, the more Covid-19 cases there are likely to be. In this sense urban form will correlate with cases of Covid-19. The study only becomes interesting if urban form correlates with the per capita distribution of Covid-19.
To elaborate, point 3 on line 260-61 concludes that ‘Areas with more epidemic distribution have higher building density’. Similarly, ln 339-340 says that ‘larger buildings have a high degree of over- 339 lap with the distribution of COVID-19 epidemic hotspots’. If we are looking at Covid-19 in absolute terms, there is nothing surprising here and we don’t need machine learning to tell us that the more people in total in a building, the more people with Covid-19 there is likely to be. However, if machine learning can tell us that the more people in a building, the more Covid-19 there will be per capita, then indeed there is a significant contribution.
Also, we should specific that we are dealing with residential buildings as Covid-19 cases are presumably registered at place of residential occupation. The difficulty, of course, is that people may have caught Covid-19 at work or in retail of other places, which does complicate matters, and this should at least be acknowledge.
One point regarding policy recommendations. If machine learning shows that building size and other aspects of urban form, affects per capita distribution of Covid-19, then, in a narrow sense, the researchers are correct in concluding that building volume and density should be reduced. However, the researchers should acknowledge that there are many other considerations apart from pandemic spread that policy makers and planners must take account of when thinking about residential density including, for example economic thresholds and vibrancy, social mix and vibrancy, urban sprawl and cost of infrastructure per capita. The impact of density and urban form is part of a mix of consideration and should not overdetermine decision making. The study is valuable for its potential contribution to the empirical basis of decision making but should not end with any dogmatic recommendation here. It is a complex field.
The paper is reasonably clear but does need a language edit as there are a small issues of grammar, and also points where clarity is a little lacking, which detract from the reading. I didn’t attempt a full edit but here are some illustrative concerns:
Ln 30-31: Rather say ‘Various studies have found…’ As currently worded, it suggests that this was the finding of your study
Ln 36-37: The following sentence needs to be revised as it does not make sense –‘They rely on crowd flow data information, on the contrary, the use of spatial information is less of the field’
Ln 48-49: Point 3 doesn’t work grammatically so I would delete a few words and simply say ‘the use of random forests to identify and analyze diseases, so as to conduct gene-wide association studies
Ln 54 – delete the words ‘which has become an important technology’
Ln 60 – I am not sure what the following sentence means: ‘In this paper, the sample of the study is improved’. Does this mean that there was a previous study and there is now an improved sample or it this an improvement on sampling techniques that other authors have used. Please clarify and perhaps indicate what other sampling techniques have been.
Ln 72-83 – Be careful of the grammar in listings. This listing for example is grammatically inconsistent and needs to be revised.
Ln 112 – Do you mean a ‘large sample size’?
p.261 – ‘The opposite is less’ – a bit confusing
Author Response
Dear Reviewer,
We have revised our manuscript according to all of your comments. Edited text in the attached revised manuscript is visible as tracked changes under the All Markup mode of Microsoft Word.
(1)Comments: The paper makes a useful contribution, indicating that most machine-based learning applications in relation to the pandemic have dealt with epidemiological prediction rather than the impact of urban environmental factors on the pandemic. The means of sampling and the deployment of machine learning is interesting and innovative. In this sense, it is a worthy paper for possible future publication.
However, I do have a major query in relation to the paper. The Covid-19 population seems to be calculated on an absolute rather than per capita basis. If this is indeed so, the value of the paper is seriously undermined. The conclusions reached in lines 256 – 261, for example, are self-evident in a way. The more people in an area, the more Covid-19 cases there are likely to be. In this sense urban form will correlate with cases of Covid-19. The study only becomes interesting if urban form correlates with the per capita distribution of Covid-19.
To elaborate, point 3 on line 260-61 concludes that ‘Areas with more epidemic distribution have higher building density’. Similarly, ln 339-340 says that ‘larger buildings have a high degree of over- 339 lap with the distribution of COVID-19 epidemic hotspots’. If we are looking at Covid-19 in absolute terms, there is nothing surprising here and we don’t need machine learning to tell us that the more people in total in a building, the more people with Covid-19 there is likely to be. However, if machine learning can tell us that the more people in a building, the more Covid-19 there will be per capita, then indeed there is a significant contribution.
Response: Yes, COVID-19 has a lot to do with population density according to traditional models. But what I want to say is that in this research, our starting point is to consider the level of urban form. The urban form is at the level of material space, and its elements mainly include roads, green spaces, water/coastline, buildings, vacant land. We add the definition of urban form and the elements of urban form considered in this paper at the beginning of Section 2（line 90-97）. In the second and third parts of the article analysis, our analysis of technical drawings and the process of calculation and deduction are also centered around these factors. In other words, the relationship between COVID-19 and urban form (material space elements) has not been revealed by further empirical research, at least in the field of urban planning or design. Therefore, our research mainly focuses on this angle. According to traditional data and models, as well as objective factual experience, most outbreaks are related to population distribution.
However, what we would like to say more is that the city we analyzed experimentally, Macau, China, it may be a special phenomenon. Macau is a high-density city. According to the 2021 census data, the population density of Macau is 20,620 people per square kilometer. After the global epidemic of COVID-19, it continued until the sudden outbreak of Macau on June 18, 2022 that we studied in this article. In the middle, there have been no new cases for more than 250 consecutive days(https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Macau). I believe this is also an interesting case. Since it is an ultra-high-density city and densely populated, why can it keep so many days without an outbreak? Therefore, we searched for the perspective of urban form (material space element) to conduct research.
(2)Comments: Also, we should specific that we are dealing with residential buildings as Covid-19 cases are presumably registered at place of residential occupation. The difficulty, of course, is that people may have caught Covid-19 at work or in retail of other places, which does complicate matters, and this should at least be acknowledge.
Response: Yes, this is a very complex question. Also limited to currently public data, many personal trips involve personal privacy. The Macau Health Bureau has not released them. Therefore, in the capture of data, to a greater extent, we obtain their residential addresses. Of course, the risk of the epidemic in the place of residence is also quite high, because from the perspective of the announced occupations, the group that started the outbreak is domestic servants, who take care of children for their employers at home, and stay in the place of residence for a relatively long time. The situation and limitations of this data acquisition, we have added in the article. (now line 116-122)
(3)Comments: One point regarding policy recommendations. If machine learning shows that building size and other aspects of urban form, affects per capita distribution of Covid-19, then, in a narrow sense, the researchers are correct in concluding that building volume and density should be reduced. However, the researchers should acknowledge that there are many other considerations apart from pandemic spread that policy makers and planners must take account of when thinking about residential density including, for example economic thresholds and vibrancy, social mix and vibrancy, urban sprawl and cost of infrastructure per capita. The impact of density and urban form is part of a mix of consideration and should not overdetermine decision making. The study is valuable for its potential contribution to the empirical basis of decision making but should not end with any dogmatic recommendation here. It is a complex field.
Response: Yes. I think the content is exactly what you pointed out. This experience is not the single determining factor. He is only used as a reference in some respects. And this reference is more based on what urban planners and architects can do in the process of practicing. We also added in the conclusion.
(4)Comments: The paper is reasonably clear but does need a language edit as there are a small issues of grammar, and also points where clarity is a little lacking, which detract from the reading. I didn’t attempt a full edit but here are some illustrative concerns:
Response: We are very grateful to the reviewer for this suggestion. Since I was not a native English speaker, I put a lot of effort into the writing process. But this effort is not particularly obvious at present, and there are still some problems with the meaning of sentences. We also made changes one by one. Please don't worry.We submitted MDPI's language proofreading and polishing services. Make sure that the passage of the article is more fluent.
(5)Comments: Ln 30-31: Rather say ‘Various studies have found…’ As currently worded, it suggests that this was the finding of your study
Response: Yes, I mean the same thing as you said. We thank the reviewer for pointing out these errors. It is indeed the research of other scholars in the past two years, not mine. Your suggestion is excellent, I have revised it for the current context.
(6)Comments: Ln 36-37: The following sentence needs to be revised as it does not make sense –‘They rely on crowd flow data information, on the contrary, the use of spatial information is less of the field’
Response: Yes. We thank the reviewer for pointing out these errors.I have revised the meaning of the expression in this sentence. The context I want to express is that traditional models rely on people flow data information for core analysis. Therefore, in the use of traditional models, material space information elements are less considered (now line 37-39).
(7)Comments: Ln 48-49: Point 3 doesn’t work grammatically so I would delete a few words and simply say ‘the use of random forests to identify and analyze diseases, so as to conduct gene-wide association studies
Response: We agree with your comments and have made some changes (now line 51).
(8)Comments: Ln 54 – delete the words ‘which has become an important technology’
Response: Yes. We thank the reviewer for pointing out these errors. This sentence has been modified according to the context (now line 58).
(9)Comments: Ln 60 – I am not sure what the following sentence means: ‘In this paper, the sample of the study is improved’. Does this mean that there was a previous study and there is now an improved sample or it this an improvement on sampling techniques that other authors have used. Please clarify and perhaps indicate what other sampling techniques have been.
Response: As you mentioned, we are the latter. The sample technique we improved in our study was aimed at other scholars in the past. The specific method of improvement is also mentioned below “First, the COVID-19 virus hotspot distribution map is used as training set A, and the city morphology map is used as training set B. The ultimate goal is to use the distribu-tion of the COVID-19 virus in cities to predict urban form, so as to deeply study the impact of urban form on the COVID-19 virus. Second, a Conditional Generative Ad-versarial Network (CGAN) is employed to analyze the relationship between urban spatial risk factors and urban form. Finally, assuming different risk distribution pat-terns, the effects of different urban patterns on the COVID-19 virus are analyzed. ” (now line 65-71)
(10)Comments: Ln 72-83 – Be careful of the grammar in listings. This listing for example is grammatically inconsistent and needs to be revised.
Response: We thank the reviewer for pointing out these errors. We thank the reviewer for pointing out these errors. At the same time, we submitted MDPI's language proofreading and polishing services. Make sure that the passage of the article is more fluent.
(11)Comments: Ln 112 – Do you mean a ‘large sample size’?
Response: Yes, what we want to say is to have a larger sample size. The number is also related to the accuracy of machine learning recognition. The current sample can make our research get a more ideal result(now line 126-131 ?).
(12)Comments: p.261 – ‘The opposite is less’ – a bit confusing
Response: You mean line 261? What we want to express here is the opposite of "Areas with more epidemic distribution have higher building density". We have made adjustments and modifications to this, and the meaning is more clear. (now line 277-278)
Reviewer 3 Report
This is a very interesting paper investigating the impact of the urban form to epidemic transmission. By applying machine learning to the urban morphology map of Macau and the trajectory data on 500 Covid-19 patients, the relationship between the Covid-19 hot spots and various urban forms is investigated. Conclusions are drawn as to what urban features (e.g. dense buildings) might increase the risk of the spread of the epidemic, and what features (e.g. road and parks) might inhibit the spread. Implications for urban planning are discussed.
Suggestion:
For readers who are not familiar with urban planning or architecture, it would be helpful to give a definition for “urban form” and which urban form elements the paper considers (e.g. roads, green spaces, water/coastline, buildings, vacant land). This can be done in the beginning of Section 2. It would also be good to add a discussion other urban-form elements that can be considered (e.g. population density, ratio of commercial/residential buildings, etc.) and how they might be incorporated to enrich the study.
Author Response
Dear Reviewer,
We have revised our manuscript according to all of your comments. Edited text in the attached revised manuscript is visible as tracked changes under the All Markup mode of Microsoft Word.
We add the definition of urban form and the elements of urban form considered in this paper at the beginning of Section 2（line 90-97）. At the same time, due to the reference to the definition concept, we have also adjusted the order of references.
Round 2
Reviewer 2 Report
The edit has significantly improved the readability of the submission, and the insertion of a few strategic paragraphs has provided the clarity that was missing on a number of issues.
