Analysis of Bluetooth RSSI for Proximity Detection of Ship Passengers

Round 1

Reviewer 1 Report

1. Please rephrase the first sentence at chapter 3, where there is no verb in the text and therefore less meaning.
2. Please correct words MARS and MRAS (lines 127 and 130), chapter 3, page 3.
3. The RSSI variation in time, for constant positions of the users, but with movement of persons in the vicinity of subject should have been also analyzed, to determine its influence on the accuracy of position determination. This is also a random component in the received signal, with importance in indoor location processes, especially for crowded places. What do you think about this scenario?
4. How do you see a practical implementation of the solution in a ship environment, where some of the separating walls may be metallic, and also there are a lot of metallic structures onboard? Is data from PACT website employed by you enough relevant for the ship scenario?
5. Please also present some ideas regarding the practical implementation of the solution: the ship will have to be equipped with beaconing BT devices, or there will be only a software application monitoring vicinity/closeness of different mobile devices and reporting the information to a control centre? Or, what other practical implementation of the solution is overseen?
6. There are too few explanations in the article if a certain model of signal propagation is taken into consideration. Overall, no signal propagation effects (other than the position of the telephone owner) are considered.
7. Is data collected from the mentioned source, in terms of quantity and quality, relevant for the ship environment conditions? For example, if two subjects are located above and behind, below 2 meters, but separated by walls, or even on the same level, but also separated by a wall – what is the solution for this false positive detection?
8. If the data is to be collected individually, shouldn’t you need also to investigate possible influence of the receiver’s position, in relationship with RSSI?
9. Overall, the work seems to be missing the mathematical modeling. Can you provide at least some equations regarding the statistical processing of data that you performed?

Author Response

Response to Reviewer 1 Comments

  Thanks very much for taking your time to review this manuscript. I really appreciate all your comments and suggestions! Please find my itemized responses in below.

Point 1:  Please rephrase the first sentence at chapter 3, where there is no verb in the text and therefore less meaning.

 Response 1: The first sentence of chapter 3 has been modified, in chapter 3, at lines 131-134, on page 3. The revision reads as follows:" The Too Close for Too Long (TC4TL) challenge, organized by the National Institute of Standards and Technology (NIST) in collaboration with the MIT PACT project, aims to improve proximity detection for Bluetooth Low Energy (BLE)-based contact tracing.

Point 2:  Please correct words MARS and MRAS (lines 127 and 130), chapter 3, page 3

Response 2: MARS has been corrected to MRAS in chapter 3, at line 135, on page 3.

Point 3:  The RSSI variation in time, for constant positions of the users, but with movement of persons in the vicinity of subject should have been also analyzed, to determine its influence on the accuracy of position determination. This is also a random component in the received signal, with importance in indoor location processes, especially for crowded places. What do you think about this scenario?

 Response 3: The impact of human movement on RSSI is mainly fading and shadowing effects. In general, the RSSI value fluctuates significantly when the human body passes between the receiver and the transmitter device. Moreover, RSSI values become more unstable. This brings considerable errors to the algorithms that use RSSI representative values for indoor positioning. In traditional indoor positioning algorithms, such as the fingerprint mapping method. This method collects RSSI data at a point for a long time and will extract a representative value to be the RSSI value of this point. The accuracy of this approach is not very high due to fading and shadowing effects. However, in the application of proximity detection, the sender and receiver are often two smartphones. The proximity detection, however, is often an example of detecting a sufficiently close distance, such as 3ft. 3ft is about 0.91m, at which distance there is rarely a human body passing by. Although the movement of the surrounding human body will have an impact on the signal, it is relatively small compared to the impact of a human body passing between the transmitter and receiver on the RSSI signal.

Point 4:  How do you see a practical implementation of the solution in a ship environment, where some of the separating walls may be metallic, and also there are a lot of metallic structures onboard? Is data from PACT website employed by you enough relevant for the ship scenario?

 Response 4: Indoor positioning must consider the impact of dividing walls on the signal, whereas proximity detection is used to discriminate between the proximity of transmitter and receiver. The purpose of proximity detection is used to find close contacts. The current transmission route of COVID-19 is still mainly droplet transmission. This means that there is no physical barrier object between the infected person and the close contact. The data from the PACT website are used to verify the feasibility of proximity detection and to improve the accuracy of the proximity detection algorithm. Although the dividing wall of the ship is made of metal, there is no separating object between the ship occupants for the proximity detection application. The data on the PACT website are collected in the absence of obstacles between the sender and receiver, a scenario that is sufficiently relevant to the ship scenario.

 Point 5:  Please also present some ideas regarding the practical implementation of the solution: the ship will have to be equipped with beaconing BT devices, or there will be only a software application monitoring vicinity/closeness of different mobile devices and reporting the information to a control centre? Or, what other practical implementation of the solution is overseen?

 Response 5: For the proximity detection application, the ship does not require additional equipment to be equipped, but only a smartphone in the hands of the ship's crew. Generally, the crew is required to download and install the application. In the cell phone, the RSSI data received from another person is used to estimate the proximity. The estimation result can be transmitted back to the control center using wifi. For privacy protection, the mobile device information of infected persons and close contacts is only known to the control center.

 Point 6:  There are too few explanations in the article if a certain model of signal propagation is taken into consideration. Overall, no signal propagation effects (other than the position of the telephone owner) are considered.

 Response 6: The signal propagation model is considered mainly because of the fading and shadowing effects. These two effects can distort RSSI signals. However, in the application of proximity detection, in general, there is no obstruction between two human bodies, and the signal does not occur as a wall reflection. The attenuation effect is mainly considered in the case of long-distance, and since the distance between two human bodies is close enough on the example of proximity detection, the attenuation effect on the RSSI signal is relatively small. Therefore, the signal propagation model is not considered as one of the main factors in this paper.

 Point 7:  Is data collected from the mentioned source, in terms of quantity and quality, relevant for the ship environment conditions? For example, if two subjects are located above and behind, below 2 meters, but separated by walls, or even on the same level, but also separated by a wall – what is the solution for this false positive detection?

 Response 7: In special cases, two human bodies are in different rooms, and a dividing wall exists between them. In addition, between different floors, there will also be a dividing wall between two human bodies up and down. Generally speaking, the Bluetooth signal will be rapidly attenuated or even lost under the blockage of the dividing wall. If the Bluetooth signal is lost, the proximity detection program will naturally stop. However, if the signal is still present, then it can be judged by the method of outlier detection. Because the RSSI strength value is very weak under the condition of wall blocking, then this weak RSSI value is considered an anomaly. If a mobile device receives weak RSSI values over a long period, then the device will not perform proximity detection. This is because, under droplet transmission conditions, there is certainly no separation between an infected person and close contact.

 Point 8:  If the data is to be collected individually, shouldn’t you need also to investigate possible influence of the receiver’s position, in relationship with RSSI?

 Response 8: This is because in the process of proximity detection, proximity is the criterion and the absolute location of the sender and receiver is not considered. Because close contacts can be identified based on the proximity detection results collected under the condition that the control center knows the mobile device information of the infected person. The mobile device information of all ship passengers is transmitted back to the control center after the application is opened. After the mobile device information of the close contacts is known, the location can be confirmed by sending messages or voice calls.

 Point 9:  Overall, the work seems to be missing the mathematical modeling. Can you provide at least some equations regarding the statistical processing of data that you performed?

 Response 9: In this paper, we first perform basic mathematical statistics on RSSI. These include meaning, median, mode, standard deviation, skewness, range, minimum, and maximum. Next, an autocorrelation analysis of RSSI is performed, a formula for which has been added in section 4.2, at lines 317-323, on page 9. Finally, the RSSI patterns were visualized using the KDE (Kernel Density Estimation) method, whose formula has been added in section 4.3, at lines 365-372, on page 10.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors present an RSSI-based detection method from the Bluetooth communications system. This approach is mentioned in the literature and widely used during the pandemic. The authors limit themselves only to demonstrating how to receive RSSI using Bluetooth, without mentioning the steps necessary to perform a conclusive analysis. The authors do not mention how they created the network of sensors for RSSI reception, what they do with the resulting statistical data, how they perform triangulation to determine the distances between passengers (RSSI of the determinant depending on the reception sensor), they only mention a Machine Learning application used. I suggest the authors to study the literature on this field and to add to the article, because the method presented is only one step in implementing such an algorithm.

Author Response

Response to Reviewer 2 Comments

  Thank you very much for your careful review and constructive suggestions with regard to our manuscript. The main corrections in the paper and the responses to the reviewer comments are as follows:

Point 1:  The authors limit themselves only to demonstrating how to receive RSSI using Bluetooth, without mentioning the steps necessary to perform a conclusive analysis.

 Response 1: In this paper, we use the public dataset of PACT for mathematical and statistical analysis, and all RSSI data have been verified and calibrated. Therefore, our analysis process is performed after the RSSI data is received by Bluetooth. We first performed basic mathematical statistics on RSSI. This includes mean, median, mode, standard deviation, skewness, range, minimum, maximum. Next, an autocorrelation analysis of the RSSI is performed, the formula for which has been added in section 4.2, at lines 317-323, on page 9. Finally, the RSSI patterns were visualized using the KDE (Kernel Density Estimation) method, whose formula has been added in Section 4.3, at lines 365-372, on page 10. The main steps to perform conclusive analysis are basic mathematical and statistical analysis, autocorrelation analysis, and KDE analysis. These steps have been described in detail in the article and the final conclusions have been drawn.

Point 2:  The authors do not mention how they created the network of sensors for RSSI reception, what they do with the resulting statistical data, how they perform triangulation to determine the distances between passengers (RSSI of the determinant depending on the reception sensor).

 Response 2: The application of proximity detection is mainly implemented based on smartphones. Therefore, multiple smartphones are able to form a sensor network. In this paper, we focus on validating the usability of proximity detection with the publicly available dataset from PACT. In the proximity detection process, distance and proximity are the main measures. The RSSI value enables to distinction the distance between the transmitter and the receiver. Therefore, this paper does not triangulate to determine the distance between passengers for proximity detection.

Point 3:  They only mention a Machine Learning application used. I suggest the authors to study the literature on this field and to add to the article, because the method presented is only one step in implementing such an algorithm.

 Response 3: Literature on the application of machine learning applied to proximity detection has been added in chapter 2, at lines 90-97, on page 2. The addition reads as follows: "Proximity detection can be achieved by indoor positioning techniques, although it can also be achieved directly by RSSI values. Applying machine learning algorithms to achieve RSSI proximity detection is a classification problem. [33] focuses on the use of RSSI data to identify whether two persons are 6 feet apart using machine learning classification. [34] studies on using machine learning approach to infer from the data collected by the sensor array to ensure classifier and a regressor on the projected distance between objects and the sensor."

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper presents an analysis of Bluetooth RSSI parameters. The authors introduced the concept of applying Bluetooth RSSI for ship’s passengers’ monitoring. The idea is very interesting, and such a solution would be useful for preventing the spreading of viruses on broad. However, the authors have not addressed the proximity detection of ship passengers. 
1. There is no connection between the authors’ analyses and proximity detection of ship passengers. The authors introduced the idea where Bluetooth RSSI can be used and then performed analysis unrelated to the proposed concept. This is because the authors used an unsuitable database for the ship environment. The database comprises the object, such as the entrance to the bathroom of an apartment - defined as a small room; the kitchen - defined as a medium room; and a large living room - defined as a large room. Considering ship, the walls and floors are made from metal, which reflects electromagnetic waves and thus creates radio shadow. Additionally, the space is restricted, and rooms are small and densely furnished. Consequently, the ship environment differs significantly from the environment used in the database. 
2. In the paper “Performance Evaluation of COVID-19 Proximity Detection Using Bluetooth LE Signal” (the paper should be described in the Related Works section), two models of the spatial behaviour of RSSI were presented. The first model is based on linear regression, while the second is based on empirical measurements. Additionally, the authors developed their method based on machine learning. In this respect, the presented analysis is a step backwards. The authors analysed the various statistics of Bluetooth RSSI measurements, which could be valuable for developing the measurement methods, but these methods have already been proposed. The authors should develop their method and verify 
3.  The purpose of the performed research is not defined. The authors should point out how the obtained results could be utilised.
4. Some results are obvious. For example, “…Using the mean value of RSSI brings less proximity detection error than median, mode, minimum 
and maximum”.
5. English language and style are not acceptable. An excellent example of a well-written article is the paper:  “Performance Evaluation of COVID-19 Proximity Detection Using Bluetooth LE Signal”.

Author Response

Response to Reviewer 3 Comments

Thanks very much for taking the time to review this manuscript. I appreciate all your comments and suggestions! Please find my itemized responses below and my revisions in the re-submitted files.

Point 1:  The paper presents an analysis of Bluetooth RSSI parameters. The authors introduced the concept of applying Bluetooth RSSI for ship’s passengers’ monitoring. The idea is very interesting, and such a solution would be useful for preventing the spreading of viruses on broad. However, the authors have not addressed the proximity detection of ship passengers.
1. There is no connection between the authors’ analyses and proximity detection of ship passengers. The authors introduced the idea where Bluetooth RSSI can be used and then performed analysis unrelated to the proposed concept. This is because the authors used an unsuitable database for the ship environment. The database comprises the object, such as the entrance to the bathroom of an apartment - defined as a small room; the kitchen - defined as a medium room; and a large living room - defined as a large room. Considering ship, the walls and floors are made from metal, which reflects electromagnetic waves and thus creates radio shadow. Additionally, the space is restricted, and rooms are small and densely furnished. Consequently, the ship environment differs significantly from the environment used in the database.

 Response 1: In most cases, Bluetooth RSSI is used in indoor positioning technology. In comparison to room size, the most critical components impacting RSSI during the development of indoor positioning algorithms are walls, metal materials, and water. The transmitter emits signals to the surrounding environment, although the signs are generally reflected due to walls and metal elements. As a result, misleading signals are generated. human mobility is everyday in typical interior placement settings, and the human body is primarily made of water, which has a significant impact on the RSSI signal. The preceding describes a typical internal placement scenario. Proximity detection, on the other hand, is not the same as traditional interior positioning scenarios. The transmitter and receiver in a proximity detection application are commonly two cellphones. The transmitter is frequently on a wall in classic indoor localization methods. A significant distance often separates the transmitter and receiver in proximity detection. The focus of proximity detection is on proximity. Two smartphone devices are pretty near to one other in practice. Proximity detection aims to discriminate between distant and near situations, not to compute distance. Significantly, unlike previous proximity detection approaches, the proximity detection discussed in this study does not add closeness via sensing distance but instead directly uses RSSI. After the distance is no longer a factor, the scenarios for proximity sensing are relatively similar, whether in a ship or a typical interior setting. As a result, the relevant conclusions gained through the dataset utilized in this work also apply to the ship environment. The answers to this comment have been added to the Introduction (Chapter 1, lines 69-88, page 2).

Point 2: In the paper “Performance Evaluation of COVID-19 Proximity Detection Using Bluetooth LE Signal” (the paper should be described in the Related Works section), two models of the spatial behaviour of RSSI were presented. The first model is based on linear regression, while the second is based on empirical measurements. Additionally, the authors developed their method based on machine learning. In this respect, the presented analysis is a step backwards. The authors analysed the various statistics of Bluetooth RSSI measurements, which could be valuable for developing the measurement methods, but these methods have already been proposed. The authors should develop their method and verify.

Response 2: Thank you so much for bringing this article to my attention. The RSSI and distance have a relatively linear relationship. Machine learning algorithms can learn this relationship when dealing with large amounts of data. The description of this article has been added in chapter 2, lines 130-134, page 3. "Performance Evaluation of COVID-19 Proximity Detection Using Bluetooth LE Signal" focuses on using an empirical model to determine the confidence. It is possible to speed up the calculation process. These are highlighted in the abstract(Original article: “As the classical estimation method utilizes RSSI characteristics models, it is faster to compute, is more explainable, and drives an analytical solution for the precision bounds proximity estimation.”), Part IV(Original article: “This approach enables faster computation in a more logically explainable manner, and it also enables us to calculate the Cramer-Rao Lower Bound (CRLB) of the performance achievable by any algorithm.”), and Conclusion(Original article: “In this paper we have presented research, development, and comparative analysis of classical estimation theory methods, which enables faster computation in a more logically explainable manner and novel hybrid model based ML approaches for proximity distance estimation using the RSSI information radiated from the broadcast channels of the BLE.”). However, in the complete text, the authors do not show the data of the graphs and tables related to the computational speed. After the confidence is obtained with the empirical model, the importance of the features is determined. Finally, three machine learning algorithms, Support Vector Machines (SVM), Random Forest, and Gradient Boosted Machines (GBM), are used to learn the relationship between RSSI and distance. It is worth noting that this article does not present a linear regression model but rather compares it with the traditional RSSI linear regression model(Original article:” In this paper, we expand the effective range of this alternative BLE specific model to about 4.5 m (15 ft) and compare the results with produced using the traditional linear regression model described by (4).”). Therefore, this article uses distance to represent proximity. However, what our paper proposes is that there is a relationship between RSSI intensity and proximity, and proximity is not defined by distance but by RSSI power. This is the most fundamental difference between our paper and this article.

This context has been added to Related works (Chapter 2, lines 134-137, page 3). As you said, the analysis that has been done in our paper is valuable and should be considered what we have already improved.

In our paper, we perform basic mathematical statistics on RSSI. These include meaning, median, mode, standard deviation, skewness, range, minimum, and maximum. Next, an autocorrelation analysis of RSSI is performed. Finally, the RSSI patterns were visualized using the KDE (Kernel Density Estimation) method. In particular, the KDE method is introduced in our paper to visualize RSSI patterns. It allows one to imagine the difference in RSSI ways near and far distances. In summary, our paper proposes a new framework for RSSI analysis to reveal the relationship between RSSI and proximity.

Point 3: The purpose of the performed research is not defined. The authors should point out how the obtained results could be utilised.

 Response 3: The purpose of our study and the role of the analysis results have been presented in the abstract. Our paper analyzes the Bluetooth RSSI signals available to the public and compares the RSSI signals into two distinct poses, stand and sit. These features can improve the accuracy and provide an essential base for creating algorithms for proximity detection. This enables the accuracy of identification of close contacts to be enhanced and COVID-19 to be stopped further within the ship. This section has been added to the Introduction (Chapter 1, lines 101-105, page 3) to make our research objectives more straightforward.

Point 4: Some results are obvious. For example, “…Using the mean value of RSSI brings less proximity detection error than median, mode, minimum
and maximum”.

 Response 4: It is important to do basic mathematical statistics on RSSI in this work. The apparent outcome is attained, and the results are declared by common sense. Our paper does not only obtain this result alone but also other exciting developments. This context has been added to the Introduction (Chapter 1, lines 98-100, page 3). If only RSSI dispersion is considered, it is difficult to reduce the error. And in most cases, the RSSI values almost belong to the left-skewed distribution. There is a combination of mean and skewness of RSSI at different poses. In the case of the close distance, the current posture of the person on board can be identified based on the mean and skewness of RSSI. As the distance increases, the difference of RSSI in various poses decreases. The mean and skewness features for distinguishing different poses will gradually disappear. In the case of the close distance, the RSSI of different poses have various fluctuation intervals in different periods. These features can improve the accuracy of proximity detection and identify the pose of the ship passengers under certain conditions. Under the condition of proximity detection with high precision, close contacts can be accurately determined, thus preventing COVID-19 from spreading further inside the ship. This context is already mentioned in Related Works (Chapter 2, lines 150-160, page 4).

 Point 5: English language and style are not acceptable. An excellent example of a well-written article is the paper: “Performance Evaluation of COVID-19 Proximity Detection Using Bluetooth LE Signal”.

 Response 5: Thank you for your advice. We have asked an English colleague to correct the details in this manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for your reply. 

Author Response

Thank you very much for your comments and suggestions.

Reviewer 2 Report

The authors tried to respond to the recommended clarifications. I have no more questions.

Author Response

Thanks reviewer for good comments and hard work. Thank you for this valuable feedback. 

Reviewer 3 Report

I appreciate the authors' detailed response to my remarks. I still have doubts about the merits of the presented work, but for sure, some amendments need to be introduced.

1. Line 21 - the conclusion that the performed analysis will contribute to the stopping of spreading the virus on the ship is too far-reaching.
2. Lines 41-45 - the Diamond Princess cruise ship passengers were localised during the one-day Taiwan trip, not aboard the ship.
3. Line 157 - what is the purpose of showing a plan of the HANNARA ship? It does not have any relevance to the paper. The authors could depict a plan of any vessel.
4. Figure 3 - the picture below presents the idea of keeping quarantined people in restricted areas, not proximity detection.
5. The equations are not visible.
6. The authors should explain how the system can work on the ship. On a ship, there is a problem with the Internet and WiFi. Consequently, how mobile phones will update their databases and exchange data? In my opinion, a router should be mounted in each room and cabin to facilitate a connection between devices.

Author Response

Response to Reviewer 3 Comments

I am very grateful to your comments for the manuscript. According with your advice, we amended the relevant part in manuscript. Some of your questions were answered below.

Point 1: Line 21 - the conclusion that the performed analysis will contribute to the stopping of spreading the virus on the ship is too far-reaching.

 Response 1: Thank you very much for your comments. We have revised Line 21 in response to this valuable review comment. Line 21 has been revised to read, "This allows for improved accuracy in identifying close contacts and can help ships sustainably manage persons on board in the post-epidemic era."

Point 2: Lines 41-45 - the Diamond Princess cruise ship passengers were localised during the one-day Taiwan trip, not aboard the ship.

Response 2: Your comment is exactly right. According to your comment, we have removed "on ships" phrase in our paper. The original sentence has been changed to read:  Using smartphone location data to track close contacts is an excellent approach to finding them (chapter 1, line 42, page 1). The reference [8] also uses mobile sensor data to track close contacts. The Bluetooth RSSI is precisely one of the mobile sensor data. The idea of our paper is to use Bluetooth RSSI from smartphones for proximity detection for close contact tracking. Since our paper is similar to the concept of this article, it is referred to in our paper. This content has been added in chapter 1, lines 46-49, and page 2.

Point 3: Line 157 - what is the purpose of showing a plan of the HANNARA ship? It does not have any relevance to the paper. The authors could depict a plan of any vessel.

 Response 3: The purpose of showing a plan of the HANNARA is that the ship environment is much similar to that of a cruise ship, and there are enough persons on board as well. In addition, the ship has many stateroom structures similar to those on a cruise ship. It is possible to demonstrate the natural application environment of proximity detection to the maximum extent. The HANNARA ship is going to be used as an experimental environment for this study in the subsequent research. This content has been added in chapter 3, lines 193-199, and page 5.

Point 4: Figure 3 - the picture below presents the idea of keeping quarantined people in restricted areas, not proximity detection.

 Response 4: According to your comment, the reason why Figure 3 is shown has been stated more clearly as follows: Figure 3 illustrates a practical application case of proximity detection. Firstly, the location information of the person riding the boat is obtained by the indoor positioning algorithm. It is assumed that the trajectory data of the COVID-19 confirmed person is known. The trajectories similar to COVID-19 are found by a clustering algorithm (DBSCAN (Density-based spatial clustering of applications with noise) as an example). and obtain its user ID to classify it as a close contact. The results of proximity detection are combined at this time to finalize the close contacts. Proximity detection can also confirm whether the close contacts are in the room. Suppose the door lock of a room has a Bluetooth signal transmitter. When the close contact leaves the room, the proximity detection is suspended as the RSSI strength changes dramatically and fades to nothing. Not obtaining the proximity makes it possible to determine whether the close contact has left the isolated room without permission. This content has been added in chapter 3, lines 210-221, and page 5.

 Point 5: The equations are not visible.

 Response 5: I'm very sorry that this problem happened. I uploaded the manuscript twice, but the formulas always failed to display for some reason. This revised manuscript is going to be uploaded in docx format and pdf version. I am sorry for this problem again.

 Point 6: The authors should explain how the system can work on the ship. On a ship, there is a problem with the Internet and WiFi. Consequently, how mobile phones will update their databases and exchange data? In my opinion, a router should be mounted in each room and cabin to facilitate a connection between devices.

 Response 6: Generally, wireless networks are installed on newly built ships. The ship's wireless network covers the active area of the ship's occupants. The RSSI data received from another person in the cell phone estimates the proximity. The estimation result can be transmitted back to the control center using wifi. The server will store the data in the local database to make it easier for the program to read the data. In older ships, a ship's wireless network has to be installed to be able to get proximity detection results from the user device and to transmit back to the control center. Furthermore, as satellite-based marine networks as the Starlink program are gradually implemented [36], the problem of data backhaul on all types of boats may be solved. This content has been added to chapter 3, lines 238-247, and page 6.

Author Response File: Author Response.pdf