Ecosystems Services and Green Infrastructure for Respiratory Health Protection: A Data Science Approach for Paraná, Brazil

Round 1

Reviewer 1 Report

The main objective this paper is ecosystem services of green infrastructure towards better respiratory health in a socioeconomic scenario typical of the Global South countries. Dataset is from national databases. May I suggest the following comments?

1. In the title of the paper, the authors mentioned “data science and modeling approach”, however, I haven’t seen any modelling inside the paper.
2. In the abstract authors mentioned “Dataset is from national databases”, it’s not clear. Need to revised this statement.
3. Dataset from just 2010? It’s about 11 years ago. I think need to explain more or involve current data.
4. It’s not clear what types of health indicators the authors involved in this study. I just saw one variable “Number of hospitalizations because of respiratory” in Table 1.
5. Data collection is cross sectional or longitudinal?
6. Please bring more evidence why we need such research?
7. Highlight the contribution of your research.
8. Hypotheses need some more explanations and sound arguments.
9. The sampling method should have explained well.
10. Missing analysis is not enough, need more explanation based on descriptive statistics.
11. Please some analysis for outliers.
12. Between lines 644-670 the authors brought some previous studies related to “asthma”. How this study related to the literature which have used in discussion section?
13. Lake of interpretation the outputs of cluster analysis.
14. Include the results of structural model in tabular form so it can be easily understandable from the readers point of view.

Author Response

Report - Reviewer 1

Authors would like to thank the reviewer. It was of great help for us to reflect deeply about our own research and improve it further. Please see below our comments on your questions. We have used the Word review tool (spell checker and track changes) and posted two files, one with the changes marked (including balloons marking to which reviewer or authors the change is due), and another “clean” with the changes included/“accepted”.

1. In the title of the paper, the authors mentioned “data science and modelling approach”, however, I haven’t seen any modelling inside the paper.

We meant modelling by the application of the CBA data-mining algorithm to learn about data set patterns. However, we agree that modelling many times involves predicting. So, we have adapted the title of the article, and changed accordingly along the paper (all marked using the Word review tool).

2. In the abstract authors mentioned “Dataset is from national databases”, it’s not clear. Need to revised this statement.

We have added “public domain”: “Data was extracted from national public domain databases”. All links are mentioned in the paper, mainly in Table 1.

3. Dataset from just 2010? It’s about 11 years ago. I think need to explain more or involve current data.

Socioeconomic issues, environmental degradation, including air quality, and health are closely associated (we have added some comments on that and some references, as well). Then, socioeconomic data and some environmental indices, apart from urban green infrastructure, such as sanitation were also included in the dataset. Then, to match timing of the variables observations, as it is a cross sectional study (we elucidated that in paper now), census data sets that also includes some environmental statistics were applied in the studies. By that, we were able to match timing of observations with both health and green infrastructure indices in the municipal spatial scale. A new census in Brazil was due in 2020. However, because of the pandemic it has been postponed. Therefore, the latest census is that from 2010. We have added some comments in the article justifying that and also included comments in the discussion of results. However, despite the limited scope, we have done some investigation using population projections in some of the data analysis. This has also been highlighted in paper.    

4. It’s not clear what types of health indicators the authors involved in this study. I just saw one variable “Number of hospitalizations because of respiratory” in Table 1.

We think perhaps some type of formatting problem occurred and the last line of the table could not be read. We did some formatting in the Table/article pages and we hope the last line can be read/seen right now. We meant RD as number of hospitalisations because of respiratory diseases divided by 100 inhabitants. We have also made sure to mention in the text the international code CID10-X. This includes a number of respiratory diseases such as asthma, bronchitis, COPD, and others. Because of the raised question number 7, we have also included some comments on how these diseases are related.

5. Data collection is cross sectional or longitudinal?

It is a cross sectional study. We have made it explicit in the paper now.

6. Please bring more evidence why we need such research?

This and the following questions (7 and 8) were treated together and we think that we were able to better explicit how this research is important> hypothesis> research question> objectives> contributions (main and lateral – other reviewers also found that this point needed some clarification). We believe that all elements are better interconnected now.  

7. Highlight the contribution of your research.

Please see the answer in question 6.

8. Hypotheses need some more explanations and sound arguments.

Please see the answer in question 6.

9. The sampling method should have explained well.

We are not sure if we understood fully what you meant by sampling method, but we thought it was about population sampling, because a question about the elderly (over 60 years old) sampling population was also raised by another reviewer. So, we have added some comments on that in the paper when presenting population numbers. Children and the elderly are more sensible to pollution and more likely to suffer from respiratory diseases (we have added some scientific evidence on that, as well). We have added some inferences, insights, about these groups (children and the elderly), mainly from the improvements we made when presenting the cluster analysis. We have also added some information about the 19 year-old threshold. This was chosen to allow health and census data to be comparable.  

10. Missing analysis is not enough, need more explanation based on descriptive statistics.

We have added some information in there. About outliers and some integrated analysis.

11. Please some analysis for outliers.

We had done that previously, but it had not been included. We have included it now, and introduced comments about outliers in the section where we presented the basic statistics and histograms, in the sections about the cluster analysis, and in the discussion of results.

12. Between lines 644-670 the authors brought some previous studies related to “asthma”. How this study related to the literature which have used in discussion section?

We have added comments and references regarding this issue in the introduction, in the research methods description, and in the discussion of results.

13. Lake of interpretation the outputs of cluster analysis.

A couple of paragraphs have been introduced in this section, and another one regarding that in the discussion of results.

14. Include the results of structural model in tabular form so it can be easily understandable from the readers point of view.

We have put up a table presenting the associative rules obtained by CBA, and improved this part in the presentation of the research methods. We think it is more comprehensible now.

 

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Although the paper addresses an interesting topic (UGI and health), this paper does not have a sound contribution to the present knowledge. The most important reason to reject this version of the paper is that whereas the title and abstract suggests that the paper is studying the relation between UGI and respiratory health, it seams that the authors have many other objectives:

 - ''the innovative approach involving machine learning'' was succesfull ...'(abstract);

• ' finding data scientific evidence .. using data science and mining'' (Introduction, suggesting that the method is part of the study)
• 'socio-economic features play an important role...'' L744
• 'lesss poverty, more sanitation, more urban ordination ... resulted in lower rates of hospitalization...'L750-L752.

And although I am not a statistician, it would have been interesting to use an integrated model with which the SEPERATE, ADDITIONAL effects of biodiversity, given that there are other factors as well that affect hospitalization (such as poverty and sanitation), had been studied. It is clear and proved with this study that they are related, but it is not clear to me, if a government would invest in sanitation for instance, being in a biodivers environment would be favourable in addition too for health. 

As a consequence of the fact that many other variables are taken into account seperately, they are presented in the Result section in a way - with much detail - that one gets confused about what the relation is between all this information and the research goal. The paper therefore lacks focus and needs serious revision in which the content reflects the title and the statistical methodology is more adequate in relation to what the authors realy want to know. 

Author Response

Report - Reviewer 2

Although we regret and do not fully agree with the refusal of the paper, we thank the reviewer and the Editor for sending us the report, which allows us to improve and make it more attractive for publishing. Below we address each point raised by the reviewer. All these have been included in the version submitted in this round. We have prepared two files, one using the Word “track changes” tools, including balloons showing to which reviewer the change is addressed, or indicating an addition proposed by the authors. The other file is “clean” as if all changes were accepted.

Although the paper addresses an interesting topic (UGI and health), this paper does not have a sound contribution to the present knowledge. The most important reason to reject this version of the paper is that whereas the title and abstract suggests that the paper is studying the relation between UGI and respiratory health, it seams that the authors have many other objectives:

• ''the innovative approach involving machine learning'' was successful ...'(abstract);

With this statement authors tried to add value to their research outcomes. We have improved the abstract to make it more comprehensible.

• ' finding data scientific evidence .. using data science and mining'' (Introduction, suggesting that the method is part of the study)

We meant that by using a proper and recognized, scientifically-accepted methodology we found “scientifically sound” results. So, research outcomes are based on observed data by applying the procedures of data science. However, we have made some inclusions to better contextualize this argument.

• 'socio-economic features play an important role...'' L744

We have now better contextualized this statement and included references to support this argument.

• 'lesss poverty, more sanitation, more urban ordination ... resulted in lower rates of hospitalization...'L750-L752.

This was mentioned in the context of presenting and discussing the knowledge acquired from the data analysis. However, the authors meant to show some of the collateral impacts of the research outcomes. We have now contextualized it better and we believe it is comprehensible.

And although I am not a statistician, it would have been interesting to use an integrated model with which the SEPERATE, ADDITIONAL effects of biodiversity, given that there are other factors as well that affect hospitalization (such as poverty and sanitation), had been studied. It is clear and proved with this study that they are related, but it is not clear to me, if a government would invest in sanitation for instance, being in a biodivers environment would be favourable in addition too for health. 

We are not sure if we fully understood these comments. The approach adopted in the paper applied statistical multivariate analysis. These included the calculation of the correlation matrix and the cluster analysis. The correlation coefficient signs shows if the correlation is positive, greater than one variable, so is the other, or, negative, one variable gets bigger as the other gets smaller. For example, this analysis revealed that the larger the “BIODIVERSITY” feature the smaller is the RD (associated with respiratory diseases). The same was verified for SAN (associated with sanitation) and low_INCOME (associated with poverty). Apart from that, “p” values measure the confidence of these correlations. The cluster analysis grouped in the same cluster RD (dependent variables), low_INCOME and BIODIVERSITY (independent variables). This means that these two independent variables are more closely related to the RD than the others.

By the comment we made about government and public policies, especially for cities of the Global South, we meant that the results showed that investments in sanitation and in the conservation of biodiversity are likely to lower hospitalisations caused by respiratory diseases. We also meant that these would have a positive impact on the economy, since less hospitalisation could reduce schooldays missed for children and reduce work absenteeism. Also, as poverty is generally entwined with health and environmental issues (we have added some comments on this and references to support that in the section we have added on research methods), by acting towards e.g. conservation and enlargements of biodiversity conservation units, that might generate “green jobs”, as well as the universalisation of sanitation. It could result in a synergic effect on reducing respiratory health issues even further, with transverse impacts on improving environmental degradation, regarding air pollution, as vegetation contributes in many cases to mitigate air pollution, and proper sanitation, among others, have impact on reducing water pollution. We made some improvements in the paper, and we think it is clearer that those are not part of the objectives of the research, but possible collateral positive impacts of it.

As a consequence of the fact that many other variables are taken into account seperately, they are presented in the Result section in a way - with much detail - that one gets confused about what the relation is between all this information and the research goal. The paper therefore lacks focus and needs serious revision in which the content reflects the title and the statistical methodology is more adequate in relation to what the authors realy want to know. 

Data science generally passes through some steps, and among them, basic statistics analysis towards the characterisation and understanding of each variable considered in the dataset/in the study. We believe these steps help data scientists to deepen their understanding about the data, which tends to help to reveal and extract knowledge from them. As we included 15 variables in the study, as a consequence a certain amount of graphical analysis is naturally one of the outcomes. We have now added more comments in the analysis of the results of the basic statistics calculations and, at the same time, we have moved some graphs to the supplementary material. We think that now the reading is flowing better and that it is more comprehensible.

Finally, apart from the changes reported in each of the reviewers’ comments, the authors have enhanced the description of research gaps, the hypothesis and objectives, articulated better the research methods to reach these goals and improved the presentation of the outcomes of the research, distinguishing them from their collateral impacts towards sustainability. We believe that after the amendments and additions we have implemented after this review have made the research reporting in the paper more comprehensible, have enhanced its scientific qualities, and have highlighted its contributions better to the MDPI Sustainability readership.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper takes an interesting approach to how ecosystem services can impact the health of residents. Chapter 1 describes the issues and challenges. It does describe in detail what kind of data was collected and in what way. However, it is unclear what the exact research question is and how the data collection contributes to it.
Especially in chapter 2, it would be useful to introduce the described method and to relate it to the goal of the paper. Table 1 shows what data was collected, but it is not clear why this variable is significant for Ecosystem Services. Figure 2-4 shows an unsurprising data set, which when put in context with Figure 5 shows that there are mainly municipalities with up to 20,000 inhabitants in rural areas. In the further course of the paper a lot of data is offered, which shows. That especially older people have a higher RD rate.

The final conclusions document known knowledge by proclaiming a higher green share. This can have a positive effect on the health of the population, although it remains unclear how this relates to the data material.

Author Response

Thank you for your insight and observations. It made us reflect about our research and promote changes to improve it. Please see below our comments on your questions. We have used the Word review tool and posted two files, one with the changes marked (including balloons demonstrating to which reviewer or authors the changes were addressed), and another “clean” with the changes included/“accepted”.

The paper takes an interesting approach to how ecosystem services can impact the health of residents. Chapter 1 describes the issues and challenges. It does describe in detail what kind of data was collected and in what way. However, it is unclear what the exact research question is and how the data collection contributes to it.

This was also mentioned by reviewer 1 and we have been able to better explicit how this research is important> hypothesis> research question> objectives> contributions (main and lateral – the other reviewer found that this needed some clarification). These are all better interconnected now.

Especially in chapter 2, it would be useful to introduce the described method and to relate it to the goal of the paper.

We have changed section 2.2 and it is now titled “Research Methods”. We have added a paragraph describing the methodology applied in the research and how it is connected with the objectives. The text also explains/justifies the following subsections, that now have been numbered 2.2.1; 2.2.2; and then 2.2.2.1; 2.2.2.2. We have also reviewed all subsections, especially section 2.2.2.2, in which we introduced more details about the CBA algorithm that was applied for mining associative rules.

Table 1 shows what data was collected, but it is not clear why this variable is significant for Ecosystem Services. Figure 2-4 shows an unsurprising data set, which when put in context with Figure 5 shows that there are mainly municipalities with up to 20,000 inhabitants in rural areas. In the further course of the paper a lot of data is offered, which shows. That especially older people have a higher RD rate.

Data was extracted from national public domain databases. The studies involved ecosystem services provided by urban green infrastructure (UGI) to protect respiratory health. The choice of data was also subject to availability in the same time scale and associated to the municipal spatial scale. Vegetation has proved to be beneficial to mitigate air pollution (e.g. Professor Nowak papers). However, although some evidence has been recently published in the literature, the direct impact of vegetation on the respiratory health remains unclear in many aspects, as some authors found the contrary. It is common sense that it depends on other factors and local conditions.

Many times, reduced green spaces are related to low cost, environmental degraded land, where the social interest population lives, especially in the cities of the Global South. This can happen in both small and large cities. Even rural municipalities can present deforested and degraded lands (lack of public policies towards agricultural land conservation). We have added some comments about these issues.

We have also added comments explaining the population segments considered in the studies and we have included some comments about the outcomes of the investigation for these population segments.

The final conclusions document known knowledge by proclaiming a higher green share. This can have a positive effect on the health of the population, although it remains unclear how this relates to the data material.

We have improved the conclusions to better explain the scientific contributions of the paper supported by research gaps, hypothesis, objectives, data material, methods, and research outcomes – thus improving the connection of these elements.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have done all my comment properly. Thank you!

Author Response

Authors thank you for revising our manuscript. We appreciated all comments. They certainly contributed to improving it. 

Reviewer 2 Report

The authors have not adequately taken the comments into consideration. In most cases, textual changes were added or phrases were changed, whereas serious concerns about focus and interpretation have been addressed. I think, for the readers of the journal, papers should be clear and sound. Although I understand that now and then contextualisation is relevant, but the authors went too far in that. 

Author Response

Authors thank you for your comments in the manuscript. We made changes in the introduction and in the discussion sections, and we believe they are better interconnected. The hypothesis and the research question have been slightly changed to make them more clear/explicit, as well as the “answers” to them, which we believe are now more directly posted in the discussion of the results. In addition to that, very minor edits were done in other sections, when needed. There was a problem on the updating and re-numbering of references, and some editing was necessary in this section as well.

Reviewer 3 Report

The whole approach how the reserach was developed can now be found in chapter 2, also the way the data was collected is now what more logical to me and gives me the opportunity to follow your approach, which can be ssen in chapter 2.2.1.

The rearrangement of the conclusion chapter has impoved the understanding of the whole paper.

Author Response

Authors thank you for revising our manuscript. We appreciated all comments. They certainly contributed to improving it.