Identifying the Daily Activity Spaces of Older Adults Living in a High-Density Urban Area: A Study Using the Smartphone-Based Global Positioning System Trajectory in Shanghai
Round 1
Reviewer 1 Report
The premise of the paper is well developed and the writing is precise with a clear flow that makes it quite an interesting read. There are a few comments or questions that I had while reading the paper.
- There are a few typos in the paper - for example, on Page 5, line 168 should be "for further processes/ processing". Although they do not hurt the flow of the paper, it is important to reduce them as much as possible.
- How did you determine the set of 80 candidates for this study? Was there a specific category (other than age and retirement status) that played a deciding factor in choosing the candidates?
- When did the 4 candidates drop out? Was it mid-experiment? If so, does their data affect any of the analyses?
- In this paper, the authors have reinforced that a concave hull provided significantly better results than a convex hull. Was this seen from this study? Is there a mathematical proof or explanation for this behavior?
- Were the IMUs in the smartphone used to determine the walking speed of the participants? IMU data can also be used in conjunction with GPS to determine points where the candidates have spent more time instead of just passing by. It can also assist in providing individual walking speeds and even patterns that can further improve the quality of such analyses in the future.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents older adults' activity spaces identified from GPS trajectory data. I understand the authors use concave hull of walking routes to represent activity spaces and analyze most visited POIs in relation to features of the built environment.
There are several strengths in this article: relatively large sample size (including long tracking period), analysis of POIs along with participant's activity spaces, and the use of comprehensive trajectory computing algorithms.
There are, however, several weaknesses that should be addressed.
First, using the concave hull of walking routes sounds like a reasonable alternative to convex hull as an activity space measure. The authors provide verbal explanation as to why concave hull is superior to convex hull, but did not provide evidences. To me, concave hull is the compromise between convex hull and network-buffer.
Second, there is little mention of how reliable it is to use the trajectory computing method (Figure 4). For example, I wonder how accurately walking segments are identified from GPS trajectories. Without some validation, it would be difficult to put faith on findings presented in this study.
Third, the current study is rather descriptive, and the authors do not attempt to render the study scientific. For instance, the authors can conduct simple statistical analysis to determine any association between walking and features of the built environment.
Fourth, there are many phrases that don't read well. Please have native English speakers read the manuscript to correct those phrases.
See below for line-by-line comments. Note that I did not make COMPLETE notes of potential English issues:
---
Line 64: visit -> visited
74-75: doesn't read well
91: health aging -> healthy aging
94: add locality to this sentence as this doesn't apply to many cities in other parts of the world
118-119: organizing meetings with neighborhoods doesn't ensure even sampling although it may help increase sample size as it depends on recruitment methods.
124: I am not sure how you determined whether recruited participants are physically active, but depending on what you mean by physical active, this (only physically active older adults make up sample) will make it difficult to generalize findings to older adults.
142: location interest -> location of interest
154: residence time -> length of residence
159 (Figure 5): describe what shade (blue to yellow) represents in Figure 5.
163: what method is used for noise filtering?
172: elaborate on spatio-temporal interpolation used to smooth outliers and missing segments of routes
208: provide some evidence that pertains to the accuracy of identifying walking routes. This is even more important as activity space measures are concave hull based on walking routes of an individual (if I am not mistaken).
221: methods for matching staypoints with POIs to derive "most visited POIs" are not described. For example, what did you do when there are multiple POIs in the staypoints location?
249: what advantage does concave hull have over network-buffer as activity space measures? The latter is more precise than concave hull. I understand concave hull (activity space measure) is created from walking segments. Does this mean that the activity space measure is not comprised of POIs that are reached by car?
258-260 (Figure 10): can you clarify on the difference between (a) and (b) in Figure 10?
294: People spent less time in Siping Science & Technology Park but they visited there frequently. Did you ask whether participants are stressed about crossing it? Or is it (feeling stressful to cross it) your interpretation? What would explain they visited there that often?
304 (46.3%): of time spent or of GPS data points?
304 (Most): report on mean and standard deviation
318 (age/gender groups): How do typical activities differ by different age groups and gender?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Authors, thank you very much for submitting a paper. After reviewing heavily reviewing your paper i came to the following conclusions:
The paper needs minor revisions.
The paper is overall of good merit but lacks some information, state of the art literature, structural refinements and some additional explanations. In the following i will display what needs to be added to the paper:
1. General Objective and Abstract: The general objectives and both the abstract is clear
2. State of the Art:
Smartphone as Data tracker: The current state of the art is missing papers that display both different usages (e.g tracking of vehicle data) of smartphone data as well as database structures afterwards. In addition some papers state different appraoches for calculating travel time. Please add them to your state of the art:
M. Reininger S. Miller Y. Zhuang and J. Cappos "A first look at vehicle data collection via smartphone sensors" 2015 IEEE Sensors Applications Symposium (SAS) apr 2015 [online] Available: http://dx.doi.org/10.1109/SAS.2015.7133607.
M. Wittmann et al., “A holistic framework for acquisition, processing and evaluation of vehicle fleet test data,” presented at the 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), Oct. 2017, doi: 10.1109/itsc.2017.8317637.
Y. Byon A. Shalaby and B. Abdulhai "Travel time collection and traffic monitoring via GPS technologies" 2006 IEEE Intelligent Transportation Systems Conference. Institute of Electrical and Electronics Engineers (IEEE) 2006 [online] Available: http://dx.doi.org/10.1109/ITSC.2006.1706820.
J. Magtoto and A. Roque "Real-time traffic data collection and dissemination from an android smartphone using proportional computation and freesim as a practical transportation system in metro manila" TENCON 2012 IEEE Region 10 Conference. Institute of Electrical and Electronics Engineers (IEEE) nov 2012 [online] Available: http://dx.doi.org/10.1109/TENCON.2012.6412332.
3. Overall structure of the paper:
The overalls tructure of the paper is good, both research design, experiments and methods are well explained and can be understood easily.
4. Results:
- Please redo figure 3, it can not be read
- Please redo figure 5 - especially the axes: i can not be read and understood
- The current results are only displayed with images. What is necessary to display the individial, quantitative results in different tables. Please add different tables that display the results to display and easy to understand overview of travel time, travel distance, mean values etc.
- One big result that is missing is the daily walking time e.g. WHEN are the most people are traveling e.g. at which hour of the day. I think this is an important results that should be correleated with the overall busy travel times in Shanghai e.g. for cars.
4. Discussion:
5. Outlook/conclusions: The current conclusions lack of an overall outlook and how this results can be used. e.g. for the development of new business modells, the integration of new transportation technologies etc. Please revise that.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
I read authors' responses to my comments thoroughly. I don't think the authors addressed my concerns adequately.
I think the most important contribution this paper could potentially make is to develop a new method of identifying activity spaces. The authors indicated in their responses that they do not focus on (A) evaluating the method for processing GPS trajectory data or (B) assessing the relationship between the built environment and health behavior (such as walking, etc.). It is clear that neither (A) nor (B) are addressed in a scientific manner as expressed by authors. Then the question goes to "does this paper really develop a 'new' method of identifying activity spaces"? Authors did not provide scientific evidence as to why concave hull can add to literature.
Authors did not respond to my questions adequately regarding:
- spatio-temporal interpolation used for outlier processing
- how concave hull is better than network buffer
- whether concave hull is derived from walking path or not
The title of the article should have some element on 'travel survey' as participants' survey or recall interview was used in identifying the daily activity spaces.
The authors can focus on (A) or (B) mentioned above to be published.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
