by Wayne M. Garrison* and Joseph H. Bochner**


Editor Note:
  With legalized OTC (over-the-counter) hearing aids sales expected as a hearing-impaired consumer option, a previous post asked if it was possible to predict hearing level without an audiogram?  This question was raised because essentially every argument against OTC hearing aid sales states that an audiogram is necessary, but is that factual?  In response to the question asked, Garrison and Bochner of NTID (National Technical Institute for the Deaf) responded, citing their extensive research stating that hearing level can be predicted without an audiogram.  They were asked if they could provide a shortened summary of their work.  They provided the article this week, and one to follow next week, that provides information on a consumer self-administered test that predicts hearing level with a high degree of accuracy.  Readers are encouraged to take the self-administered online test https://apps.ntid.rit.edu/NSRT/.  


Accommodating the OTC Hearing Aid Discussion: Hearing Screening in the Absence of Audiometry

From retail politics to humanitarian outreach efforts to expand delivery of hearing healthcare services, the OTC Hearing Aid Act of 2017, introduced in the 115th Congress of the USA on March 21st of this year, garners reactions ranging from anxiety to excitement. Chief among the concerns expressed by audiologists is the obvious absence of policy around the requirement in the OTC discussion for an audiogram to guide consumers in the purchase of an appropriate hearing aid.  Policymakers are, nonetheless, spurred on by the need to expand the market of those not currently served with hearing aids, pointing to accessibility and affordability as major barriers. Is there a middle ground within which the interests of policymakers, audiologists and consumers might be better served? In this article, we introduce an online hearing screening test intended to accomplish that end, in essence bringing a potential profession/legislative divide together through an individualized assessment component that ultimately benefits the end user, a person with hearing loss.


Self-Testing of Hearing

Contemporary approaches to automated hearing screening in adults typically involve some form of self-administered test, with stimuli often presented under variable and uncontrolled listening conditions. The tests typically are delivered to individuals via land-line or cellular telephones, the internet, or hand-held consumer electronic devices such as smartphones and tablet computers. Although software applications have been developed for determining hearing thresholds and screening with the use of pure tones, they have proven problematic for even the most promising automated tests.

Partly because calibration problems make it difficult to deliver pure-tone tests over the Internet and on wireless devices, stimulus materials consisting of speech, more often speech in noise, are used in some of the newer hearing screening tests. Speech recognition testing has a long history involving a variety of tests, stimulus materials and administration protocols.  Speech-in-noise screening tests provide a good measure of functional hearing capabilities, but their relationship to pure-tone testing results, until now, has been quite limited. Seeking to reverse that trend, recent research at the National Technical Institute for the Deaf1,2 (NTID) has resulted in the development of a self-administered automated speech-based hearing screening test (NSRT®) enabling the prediction of hearing thresholds with impressive accuracy.

 

Self-Administered Automated Speech-Based Hearing Screening Test (NSRT®)

Data obtained from the NSRT® testing experience are used to construct a pseudo audiogram. In their work the researchers report that, when predicted hearing thresholds are compared with conventional, clinical pure-tone measures, the sensitivity and specificity of the NSRT® screening measure have been found to be 95% and 87%, respectively; diagnostic accuracy is 91%.  These statistics reinforce the argument for OTC hearing aid sales suggesting that, in addition to self-fitting, self-assessment may be a viable way to manage the market of those individuals not served by current audiological practice.


The NTID Speech Recognition Test (NSRT®) Described

The NSRT® is a computerized, adaptive hearing screening test.  It is composed of sentence-length stimulus materials containing phonetic contrasts, primarily minimal pairs. The test simply requires respondents to determine whether sentences printed on a computer screen are the same or different from sentences delivered as auditory stimuli through the computer sound card. Respondents are encouraged to take the test using headphones.

As test takers respond to tasks presented in an adaptive test, the test adjusts itself by selecting the next stimuli to be presented on the basis of performance on preceding tasks.  For example, in the assessment of academic performance in a given domain, if a test taker performs well on a set of intermediate-level tasks, more difficult tasks are presented.  Conversely, if a respondent performs poorly on intermediate-level tasks, tasks of lesser complexity are presented. In an adaptive testing situation, testing terminates when the performance of a test taker on a trait or construct reaches its highest sustainable level or threshold.

In adaptive testing, test takers respond to tests composed of different stimulus materials. The psychometric technology that allows equitable measures to be determined across differing configurations of stimulus test materials is item response theory or IRT, the preferred methodology in the field of psychometrics for optimizing statistical information yield available for test takers.

Like the stimuli used in other psychophysical procedures, the stimuli used in the NSRT® are scaled along a continuum extending from low to high degrees of magnitude.  However, rather than representing a physical construct such as the intensity of a sound, the continuum in this instance represents a domain of human performance, speech recognition ability. The scale values separating the NSRT® stimulus materials are the product of previous research undertaken by the authors but, briefly explained, represent differences in the complexity of discerning linguistic contrasts between stimuli in the testing protocol (minimal pairs), which are themselves associated with variation in the phonetic and acoustic properties of speech. 

 

The Hearing Screening Interface

To access the NSRT® hearing screening test, go to https://apps.ntid.rit.edu/NSRT/. There is no cost to take the test; it is freely available. All of the information provided by individual users is confidential and will not be shared for any purposes, commercial or otherwise.

Initially, individuals who wish to take the hearing screening test are prompted to create an account.  This is accomplished by providing an email address and a password of one’s own choosing. Individuals who have previously been tested are simply asked to log on to access earlier test results or to re-test under the same/different condition (i.e., quiet vs. noise).

Figure 1. An NSRT® practice item. Was what you heard the same or different from what is written? The person taking the test uses the mouse to click on “same” or “different.” If the person taking the test does not believe the test stimuli is loud enough, they are encouraged to adjust the computer’s volume control.

Users are next asked to respond to a brief set of questionnaire items (7) intended to gauge their perception whether they suffer from hearing impairment. They are next instructed how to adjust the listening level of their computer for administration of the hearing screening test. A practice test, which returning users can bypass, is provided to familiarize respondents with the nature of the stimulus materials and testing procedure. An example frame corresponding to the presentation of a practice item is shown in Figure 1.

Responses to the discrimination tasks in the practice sequence include feedback (i.e., correct/incorrect). Users can adjust the volume of auditory signals during the practice session as well.  The  icon in the upper right corner of Figure 1 alerts respondents that auditory signals (i.e., spoken sentences) are being presented.

Individuals visiting the assessment site who are unable to correctly discern whether the printed and auditory stimuli are the same/different on fewer than half of the practice items (i.e., relatively “easy” discrimination tasks) are not advanced to the formal testing stage.  They are advised that, given the data that they have provided to this point, further testing would likely provide questionable results. These individuals have hearing sensitivity outside the area targeted by the NSRT®. This is further corroborated by analysis of responses to the questionnaire items.  Individuals who “pass” the practice test advance to the formal hearing screening test.

 

Summary

To summarize, persons visiting the hearing screening test site are asked to: (1) create an account; (2) provide some background information via a brief questionnaire; (3) set the volume of their computer or wireless electronic device to MCL; (4) take a brief practice test to familiarize them with the nature of the assessment task; and (5) take a hearing screening test, the results of which are used to create a pseudo audiogram. Thereafter, respondents are provided with a variety of informative reports regarding their test performance.

Continuation:  The information reports and interpretation resulting from this testing will be provided in next week’s post. Readers are encouraged to take this online test at https://apps.ntid.rit.edu/NSRT/ to more fully understand and appreciate this self-test of hearing.


References

  1. Bochner, J. H., Garrison, W. M. and Doherty, K. A. (2015).  The NTID Speech Recognition Test:  NSRT®. International Journal of Audiology, 54, 490-498.
  2. Garrison, W. M. and Bochner, J. H. (2015).  Applications of the NTID Speech Recognition Test (NSRT®). International Journal of Audiology, 54, 828-837.

 

*Dr. Wayne Garrison is a Research Professor at the National Technical Institute for the Deaf on the Rochester Institute of Technology campus in Rochester, New York.  He is a psychologist by training, with a broad range of R/D experience in statistics, psychometrics and software design.

**Joe Bochner is a professor and department chair at RIT/NTID.  He has been involved in the language sciences, deafness and higher education for four decades, conducting research on the acquisition of English language and literacy skills, speech perception and production, and American Sign Language.  

It is interesting that although many products have been designed to be worn in the ear, there is little information specifically about the ear concha dimensions of the outer ear, the most likely location for placement and retention.  This would seem to be of much more significance than many other ear measurements that have been made, especially for those who design products that are ear-worn and intended for the general consumer market, such as OTC (over-the-counter) hearing aids, earplugs, hearables, vital signs monitoring ear units, earbuds for acoustic listening, communications systems, etc.

A search shows that there is not much data on concha measurements.  And, that of which there is, has been measured in a number of ways that makes comparisons difficult.  It is likely that designers of in-ear products may have captured concha measurement data themselves, but when designing proprietary products, such data is not generally shared.  Therefore, this post provides what published concha measurement data that could be found using a reasonable search time period.


Concha Depth

Figure 1. A method for concha depth determination. The depth should be measured to the deepest part of the concha on an angle orthogonal to a line connecting the tragus anteriorly and the outermost helix contact posteriorly.

Providing “real estate” for in-ear products is one of the most immediate critical mechanical requirements.  A most popular method to make this measurement is shown in Figure 1.  Figure 2 shows results of concha depth measurements for adult ears as reported by different researchers, with concha depth ranging from 9 mm to 16.8 mm.  Unfortunately, measurement methods were not always provided, and the number of subjects varied considerably, both of which should be recognized when evaluating these data.  Of the data in Figure 2, that by Burkhard & Sachs, Algazi, and Staab used the concha depth measurement identified in Figure 1.  It was not clear how the other concha depth measurements were made.

Figure 2. Concha depth measurements in mm as reported in the anthropometric literature. The range is quite high, definitely reflecting differences in the methods used to record concha depth.


Concha Width

Concha width is measured in different ways among the articles reported in this post, as shown in Figure 3.  For this reason, concha width data cannot be compared directly, even though different methods could produce similar results, depending on the shape of the concha.  The only consistent measurement point seems to be that the measurement is made to the most posterior expansion of the concha.  The anterior measurement landmark reference varies among the methods, as does how the distance between the reference points is measured.

Figure 3. Concha width measurement variations as identified in the references in this post. The red dots identify the anterior and posterior landmarks of the measurement.

Concha average width in mm is shown in Figure 4 for combined genders.

Figure 4. Concha width average measurements in mm for male and female ears. In all the studies identified in this graph, the genders have been combined, with the exception of where certain ethnic group genders have been identified.

Concha Length

Concha length measurements do not show significant differences in measurement.  This is ordinarily described as being from the base of the intertragal notch to the maximum height of the concha cavum.  Exceptions generally relate to adding the concha cavum height to the concha cymba height.  However, a major departure is shown in Figure 5, identified by the orange bar.  In this case, the concha height is described as measured from the base of the intertragal notch to the upper part of the concha cavum.  In other words, the concha cymba was not considered to be included in the concha height dimension.  Research data suggests that different ethnic populations have different sized conchae, but Figure 5 shows the range to be less than 4 mm.

Figure 5. Concha length (height) average measurements in mm for male and female ears. In all the studies identified in this graph, the genders have been combined, with the exception of where certain ethnic groups have been identified. The orange bar considered the concha height to extend from the base of the intertragal notch to the upper part of the concha cavum.


Gender Ear Concha Length and Width Comparisons

Essentially every measurement study of the auricle shows that the male ear is larger than the female ear. The same generalization holds true for concha length/height and width measurements, as shown in Figure 6.  (The Bozkir et. al. data was confusing in the reference used, and should be rechecked because it seems to be at odds with all the other gender concha width comparisons).

Figure 6. Gender concha length/height and width differences in mm as reported in the literature. This comparison is based on studies that showed both genders’ length and width measurements.


Summary

Consumer and medical products are moving toward in-ear placement.  Because of this, it would appear that ear measurements, primarily concha height, width, depth, volume, and configurations would be high on the list of important data.  This is especially true for those products that are intended to fit multiple-sized conchae with relatively generic sizes(s).


References

Algazi, V.R, Duda, R.O., Thompson, D.M., Avendano, C. (2001).  The CIPIC HRTF Database, IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, 2001.

Bozkir MG, Karakas P, Yavuz M, Dere F (2006). Morphometry of the external ear in our adult population, Aesth. Plast. Surg., 30(1): 81-85.

Burkard MD, Sachs RM.  (1975).  Anthropometric manikin for acoustic research, Journal of the Acoustical Society of America, 58, 214-222.

Fels, J.  From children to adults: how binaural cues and ear canal impedances grow. January 30, 2008. Von der Fakula ̈t fu ̈r Elektrotechnik und Informationstechnik der Rheinisch-Westfa ̈lischen Technischen Hochschule Aachen zur Erlangung des akademischen Grades einer DOKTORIN DER INGENIEURWISSENSCHAFTEN genehmigte Dissertation.

IEC-Publication 126 (1973).  IEC reference coupler for the measurement of hearing aids using earphones coupled to the ear by means of ear inserts.  Bureau Center. De la Comm.  Electrotech. Int. Geneve, Suisse. Second edition.

Kapil V, Bhawana J, and Vikas K. (2014).  Morphological variation of ear for individual identification in forensic cases: a study of an Indian population.  Research Journal of Forensic Sciences, Vol. 2(1), 1-8, January.

Natekar PE, De Souza FM. Demarking and identifying points-reliable criteria for determination of sex from external ear. Indian J Otol 2012;18:24-27.

Sibbald, A. (2000).  Virtual ear technology.  Sensura Limited.  Company publication.

Singh P,  Purkit R. (2006).  Anthropological study of human auricle.  JIAFM, 28(2); 0971-0973.

Staab, W. (2002).  Concha depth.  Unpublished.

Stavrakos SK, and Ahmed-Kristensen A. (2012).  Assessment of anthropometric methods in headset design.  International Design Conference, Dubrovnik, Croatia, May 21-24.

Teranishi R, Shaw EAG. (1968). External-ear acoustic models with simple geometry. J Acoust Soc Am 44 :257-263.

Verma P, Sandhu HK, Verma KG, Goyal S, Sudan M. Ladgotra A. (2016).  Morphological variations and biometrics of ear: an aid to personal identification.  Journal of Clinical and Diagnostic Research. Vol. 10(5) 238-142.

Zwislocki J.  (1970).  An acoustic coupler for earphone calibration Rep. LSC-S-7.  Laboratory of Sensory Communication, Syracuse University, Sept.