Non-REM Sleep Marker for Wearable Monitoring: Power Concentration of Respiratory Heart Rate Fluctuation
, Norihiro Ueda 1, Masaya Kisohara 1, Yutaka Yoshida 2, Haruhito Tanaka 3 and Emi Yuda 4
Mohammad Imtiaz
Round 1
Reviewer 1 Report
Comparison to polysomnography is not correct as the epoch used for Hsi index computation is not 30s. If 30s epoch is used for the computation of Hsi index then the AUC of Receiver operating characteristics may drop down drastically. Recommended that the authors compute AUC for 30s epoch and report the results in the revised manuscript.
Author Response
Comparison to polysomnography is not correct as the epoch used for Hsi index computation is not 30s. If 30s epoch is used for the computation of Hsi index then the AUC of Receiver operating characteristics may drop down drastically. Recommended that the authors compute AUC for 30s epoch and report the results in the revised manuscript.
As indicated by the reviewer, the discriminant analysis was performed for both the dominant sleep stage per 5-min segment and the sleep stage per 30-sec epoch. (line 143). Table 4 was added to present the results (line 176). We discussed these results in the limitation section (line 247).
Reviewer 2 Report
Non-REM Sleep Marker for Wearable Monitoring: Power Concentration of Respiratory Heart Rate Fluctuation
Junichiro Hayano et. al.
This study presents a novel method for estimating non-REM sleep from heartbeat signals.
The new bio-signal marker (Hsi) has an advantage over existing methods because it relies on a single input (e.g. ECG). Its univariate analysis is comparable to other methods which use multiple variables. This is clearly an advantage, especially in wearable implementations.
Methodology and analysis used in this study are accurate. The experimental data and analysis support the conclusions, and the logic of the paper is sound.
The manuscript is well written and succinctly presented.
It would have been interesting to see a scatter plot of Sleep Stage versus Hsi % in Figure 3. Hopefully three clusters would be visible with REM separated from the other two.
Minor corrections:
Line 18: change “bases” to “basis”
Define w (omega) explicitly before use in section 2.1
Correct Figure 2 caption;
line 233: change “has not taken sleep” to “not slept”
Comments for author File: 
Comments.pdf
Author Response
It would have been interesting to see a scatter plot of Sleep Stage versus Hsi % in Figure 3. Hopefully, three clusters would be visible with REM separated from the other two.
>> Since sleep stages are discrete categories and not continuous values, we could not effectively present three clusters of Hsi values corresponding to sleep stages. Therefore, instead, Figure 4 was revised to show the exact distribution of Hsi and BM values associated with sleep stages.
Minor corrections:
Line 18: change “bases” to “basis”
>> We corrected the error (line 18).
Define w (omega) explicitly before use in section 2.1
>> We defined omega explicitly (lines 65 and 74 and in Figure 2 caption).
Correct Figure 2 caption;
>> We revised Figure 2 caption.
line 233: change “has not taken sleep” to “not slept”
>> We corrected the error (line 239).
Thank you!
