Assessing Drivers’ Physiological Responses Using Consumer Grade Devices
Round 1
Reviewer 1 Report
This is a work that focuses on an interesting topic. How can I measure physiological responses with wearable devices with good quality?
The conclusions are interest and are supported by the results. Nevertheless, I have some questions regarding the data quality.
The paper needs some writing review. The are expressions not usual, which makes the reading difficult.
On results section, I suggest that you systematize the results on a comparison table complementing the text. In my opinion, it will be easiest to evaluate the differences between conditions.
In my opinion authors should avoid mention other sections in a different one. For example: mention the “Materials and Methods” on Introduction.
On related work, it should also be presented other works on assessment of drivers physiological response and not only on the equipment and devices used.
Some references are not provided, in the manuscript several times appears the reference marked as error.
Why are you only evaluating the HRV, and not other ECG features?
When you described the Faros, I did not understand if it provides the ECG or only features from the ECG.
When you describe the tasks (line 291). Is it enough 5 minutes of data? 5 minutes are enough to induce effects on driving performance evaluated by physiological signals? Can this be a reason for the results, when you do not find significant differences between the two driving tasks?
On the results (lines 379-381), you mention that there are missing values on the E4. Which are the percentage of missing values? You do not have access to the ECG?
In the Mean HRV section, lines 383-387, is it possible that the missing values have influence on the results?
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
the reviewer was wondering are the E4 and Faros 360 are designed to be used to take measurements under driving conditions. The researcher used a driving simulator, the high fidelity of this type of driving simulator needs to be verified. the drawback of using a driving simulator is the participants know it is not a real driving situation, they wouldn’t have any safety concerns, this will definitely affect the measurements. There is no information on the driving simulator scenarios segment length, travel time, roadway geometry. Seems to the reviewer that there are not many conflicts points and safety risks, but actually, the high demanding situation is more interesting to transportation research than free flow driving condition. Seems to the reviewer that the underlying assumption of the statistical analysis is the measurements from the driving condition are different from those taken under driving conditions. This assumption needs proof. For this free flow driving condition, the ground truth could be it is not statistically different from the baseline no-driving condition proof reading is needed. There are mis-links of tables and figures.Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper is improved, and the questions answered by the authors.
On the Reviewers' Comments and Authors' Responses:
Comment R.1.6 Why are you only evaluating the HRV, and not other ECG features?
In the E4 site, they report that the sensor measure blood volume pulse, that is different from blood volume pressure.
Comment R.1.7 When you described the Faros, I did not understand if it provides the ECG or only features from the ECG.
Since Faros provide raw ECG, why did you not process it?
Comment R.1.8 When you describe the tasks (line 291). Is it enough 5 minutes of data? 5 minutes are enough to induce effects on driving performance evaluated by physiological signals? Can this be a reason for the results, when you do not find significant differences between the two driving tasks?
I agree with the 5 minute processing of the ECG, considering the literature. However, you can design a longer task, and divide the data in 5-minute window.
On the Revised Paper:
Page 9, lines 307-308: The HRV evaluation should be done on 5-minute segment. However, you can collect longer ECG records and evaluate 5-minute window only.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
