Towards Auditory Profile-Based Hearing-Aid Fitting: Fitting Rationale and Pilot Evaluation

Round 1

Reviewer 1 Report

This is a manuscript exploring new methods of hearing aid fitting. It think that the manuscript is innovative and interesting and merit publication.

Author Response

Thank you for your comments and for reviewing the current version of the manuscript.

Reviewer 2 Report

The manuscript from Sanchez-Lopez et. al reports the findings of a pilot study that explored the potential for employing auditory profiles and stratification for hearing-aid fitting. The study is of significance and will be of interested to the readers of Audiology Research. The approach, design of the study and presentation of the results are sound. Overall, the manuscript is well written. I only have some minor comments.

1) It’s stated in line 114 that profile B group had only 1 participant. “Profile-B listeners”  in lines 144 and 149 should read “Profile-B listener”.

2) It’s not clear how the mean preference ratings presented in figures 3 and 4 were calculated. Are they the mean ratings of the participants or the mean of multiple trials? If it’s the former, how was the standard deviation calculated for profile-B group?

3) This reviewer is of the opinion that a mixed-model approach would have been superior for statistical analysis (as elaborated in chapter 6 of reference #15).

Author Response

Thank you for your comments and suggestions and for your time reviewing the current version of the manuscript

1) It’s stated in line 114 that profile B group had only 1 participant. “Profile-B listeners”  in lines 144 and 149 should read “Profile-B listener”.

Thank you for pointing this out. The text has been modified accordingly.

2) It’s not clear how the mean preference ratings presented in figures 3 and 4 were calculated. Are they the mean ratings of the participants or the mean of multiple trials? If it’s the former, how was the standard deviation calculated for profile-B group?

The preference ratings presented in figures 3 and 4 are the mean of 3 repetitions for profile-B in quiet (figure 3) and the 3 repetitions in 2 sound scenarios (traffic and cafeteria) for profile-B in noisy sound scenarios (figure 4). For profiles A, C and D, it is also the averaged across listeners (2 in each group). 

The text has been modified making this point clear in the captions of Figures 3 

"The circles represent the mean of the scores of the three repetitions performed by the listeners belonging to profile [...]"

and Figure 4:

"Each circle represents the mean of the scores of the three repetitions performed in each noisy soundscenario (traffic and cafeteria) by the listeners belonging to profile [...]"

3) This reviewer is of the opinion that a mixed-model approach would have been superior for statistical analysis (as elaborated in chapter 6 of reference #15).

Thank you for your comment. We agree with the reviewer that doing the analysis using mixed linear models would be more adequate even if the sample size is small.

The analysis has been repeated using a linear mixed model described now in Appendix A. The model is the result of applying the step function of the lmer package in R which performs "backward elimination of random-effect terms followed by backward elimination of fixed-effect terms in linear mixed models."

The model: 
Rating ~ Profile + Condition + Background + HAS + Profile x HAS + HAS x Background + "Subject"

Being "Subject" a random effect nested with Profile. Although the complete model included the repetition as a random effect, this was found not significant.

Figures 3 and 4 now show the honest significant differences result from the linear mixed model analysis and the data repository has been updated with the R code of the analysis. Furthermore, the analysis of the pairwise differences between the three Background supports the division of the results in a figure containing the Quiet scenario and one containing both noisy environments. 

The code used for this analysis has been added to the Zenodo repository containing the data and materials of the study:
http://doi.org/10.5281/zenodo.4421553 

Reviewer 3 Report

The subject of the article is extremely important and actual. The topics is really hot.

I have the following comments and questions for the authors. There are many awkward phrases that I do not point out here; I only point out those where the meaning cannot be interpreted:

The figure 2 is extremely hard to read.

The paragraph between 175-194 is not clear can be rewrite in more clear format.

The conclusion need to clear and specific. My recommendation is to focus on short conclusion.

Please recheck the References order.

Please double check the article by a native English reader.

Thanks again for the chance of reading the article.

Author Response

Thank you for your comments and suggestions. 

1) There are many awkward phrases that I do not point out here; I only point out those where the meaning cannot be interpreted:

The manuscript has been thoroughly revised to improve its clarity.

2 ) The figure 2 is extremely hard to read.

Thank you for your comment. The Figure has been modified reducing the text and complexity. 

3) The paragraph between 175-194 is not clear can be rewrite in more clear format.

Thank you for your comment. The subsection "5.1. Gain prescription" has been rewritten as:

"The gain prescription could be prescribed by different formulas in different sub-populations based on their hearing deficits. Here, the proposed fitting formula applied different correction factors to an audiogram-based nonlinear gain prescription (Expression 1). This simple strategy allowed for gain adjustments that prioritised either audibility maximization or loudness normalization based on comparisons with other gain prescriptions (Table 1). However, current fitting formulas, for example NAL-NL2 \cite{Keidser2011NAL-NL2}, prescribe the HA gain based on an optimization process and a trade-off between two models: a speech intelligibility model and a loudness model. Besides, it consider gender differences in terms of amplification preference, binaural summation and HA user experience in the gain prescription. Although the complexity of NAL-NL's prescription rule might appear more individualized, it does not directly consider the listener's supra-threshold hearing abilities in the models nor measures beyond the pure-tone audiometry.

The results of the pilot evaluation suggested that listeners in Profiles C and D may benefit from a prescription aiming for loudness normalization in opposition to NAL-NL2, which is based on audibility maximization. However, the approach of the NAL-NL2 fitting formula may be revised to allow a more personalized HA fitting. Regarding the findings from [Sanchez-Lopez et al. (2020)] and the present study, the models used in the NAL-NL2 prescription may be modified to include SI-related deficits and LP-related deficits as input parameters. This can be implemented in the clinic by using the results of loudness scaling tests needed for loudness restoration [(Oetting et al. 2018)] and speech-in-noise perception tests. Consequently, the prescription would apply different criteria depending on the auditory profile of the listener, and the optimization process would provide a weighted solution where either speech audibility or loudness are prioritized."

4) The conclusion need to clear and specific. My recommendation is to focus on short conclusion.

Thank you for your comment. The conclusion has been rewritten as:

"The present study proposed an auditory profile-based fitting rationale for more stratified HA treatment. The results of a pilot evaluation showed small but significant differences in terms of preferred HA setting among listeners belonging to four distinct profiles. Overall, these initial findings provide a useful basis for further investigations into profile-based HA fitting, which will include field trials with wearable devices and objective assessments such as speech intelligibility tests."

5) Please recheck the References order.

We have rechecked the references order but we did not find any inconsistency.

6) Please double check the article by a native English reader.

Thank you for your suggestion, the final version has been revised by a native English speaker.