Selecting the Optimal Signal Processing for Your Patient

Author: Lynn Sirow, Ph.D. & Pamela Souza, Ph.D.

Let’s say that you are consulting a cardiologist for treatment of heart disease. After a careful examination, he prescribes medication for your condition. When you question him about the specific medication, he says, “Try Lasix for a week. If it doesn’t work, then try propranolol for a week, and see which one seems to work better for you.” Perplexed, you ask him, “But which drug is best to treat my heart condition?” He responds to you by saying, “Well, whichever drug you like will be best, and if you don’t return to my office to complain about it, I’ll assume things are fine.”

For most of us this scenario seems outrageous because we expect a cardiologist to rely on evidence-based practice to make treatment decisions. Now, consider the selection process many audiologists use to dispense hearing aids. Commonly, audiologists select a product line based on familiarity with the programming software, confidence in the sales representative, or pricing (Johnson, 2007; 2011; Johnson et al. 2009). The choice of manufacturers and/or product is often made without considering how different signal processing approaches can affect the patient’s speech understanding ability. Finally, hearing aid outcomes may not be formally measured. For many audiologists, the fitting is considered successful if the patient does not return asking for a refund within the trial period.
Algorithms Vary Across Manufacturers
Although there are similarities, each manufacturer uses different sound processing. For example, if you program the same audiometric information from a single patient into several manufacturers' fitting software, you are very likely to have several different frequency responses, compression ratios and speeds, and a host of other differences across these manufacturers. They all can’t be the optimal fit for your patient. While much has been written on the most accurate way to program a hearing aid (Taylor & Mueller 2011), there has been little research on how to choose the appropriate algorithm (or manufacturer) that is best for each patient. We attempt to shed some light on this process by looking at one important feature found in all modern hearing aids.

This feature, which varies in implementation across manufacturers, is compression speed. Compression speed is typically quantified by the hearing aid’s attack and release times across individual channels of signal processing. As the data in Table 1 suggest, there are substantial differences in release times across major manufacturers. Several laboratory studies (e.g., Gatehouse et al. 2006; Lunner & Sundewall-Thoren, 2007) have shown that patients with higher scores on a cognitive test that examines the ability to process and store information (i.e., working memory) perform better with fast attack and release times, and those with poorer cognitive abilities perform better with slow attack and release times. Gatehouse et al. (2006) also linked benefit from fast-acting compression to an individual’s temporal and spectral resolution. While these findings have received considerable attention in the published research, they have not been implemented into audiologists' clinical decision-making processes. While some clinicians may be unaware of these research findings, others may be unsure of how research from the laboratory on compression speed can be incorporated into the clinical decision-making process. This paper presents one possible approach to incorporating such laboratory-based work into patient care. We evaluated two different pre-fitting measures and their relationship to compression speed (i.e., release time of compression) for a group of patients fitted in a private practice clinic.
Table 1: Examples of attack and release times reported by manufacturers. Note the wide variety of release times
Attack (ms) Release (ms)
Acto (Calm) 5 1000
Acto (Gradual) 10 640
Acto (Active) 10 320
Dynamic 10 160
Energetic 10 80
Audeo 1 50
Motion (Dual) 5 900
Motion (Syllabic) 10 100
Ino 5 25
Sonic Innovations
Flip 2 12
Ion 10 12
Moxi 20 8-15 30-350
Clear 440 10 ms – 2 sec 20 ms – 20 sec
Is there an optimal aid for every patient?
In this study, each participant was fitted binaurally with different pairs of behind-the-ear hearing aids. The hearing aids selected for a participant represented a range of compression speeds from slow to fast, using products that might be prescribed for the patient’s needs and preferences. For example, the fitting audiologist considered ease of use such as rechargeable batteries in cases of limited dexterity or availability of a manual volume control. Speech-in-noise recognition (QuickSIN; Killion et al., 2004) was measured in soundfield for each pair of aids. Results, which are shown in Table 2, indicated that 17 of the 27 subjects had a significant difference at the 95% confidence level in their aided SNR loss score between the fast-acting and slow-acting compression release times. These data suggest that compression release time has some effect on aided speech recognition ability as measured using the QuickSIN. In addition, informally collected data from this group suggested that the aid that “sounded the best” during the clinic appointment was not always the aid that provided the best aided SNR loss score. Although patient perception of sound quality is important, we would argue that the first priority for most patients is improving aided communication in background noise. This is particularly true given that perception of sound quality may improve over time as the patient adapts to the amplified signal (e.g., Convery & Keidser 2011).
Table 2: QuickSIN test SNR scores: the score difference between fast-acting and slow-acting compression (scores based on two lists)
Patient Difference (SNR) Patient Difference (SNR)
#9 7 #21 3.5
#17 5.0 #22 3.5
#23 5.0 #19 3
#2 4.5 #25 3.0
#3 4.5 #6 2.5
#8 4.5 #5 2
#24 4.5 #13 2
#26 4.5 #14 2
#27 4.5 #15 2
#1 4.0 #16 2.0
#4 4 #12 1.5
#7 3.5 #18 1.5
#10 3.5 #11 1
Significant Differences: At 95% confidence interval: 2.7 dB
90% confidence interval: 2.2 dB
80% confidence interval: 1.8 dB*
*Etymotic Research Manual for the Quick SIN test, p. 20, 2004.

Is there a test that helps select the optimal release time?
One of the primary objectives of the pre-fitting appointment is gathering objective information about the status of the individual’s auditory system that can be used in the selection of the most appropriate hearing aid. Historically, several tests, including speech-in-noise tests, have been used in the test booth to try to accomplish this goal. Currently, the most popular pre-fitting measure used for making amplification decisions is the pure-tone audiogram. The audiogram may direct the choice of hearing-aid gain, compression threshold and compression ratio across frequencies, but it doesn’t help choose the appropriate compression speeds.

Fortunately, tests of working memory may provide such guidance on the selection of the optimal compression release time for each patient. Souza and Sirow (2012) employed a test of working memory, the Reading Span Test (RST), used in work by Lunner and colleagues (e.g., Lunner & Sundewall-Thoren 2007). In that test, patients are asked to judge whether sentences presented on a computer screen make semantic sense, then patients are asked to recall portions of the sentences. This processing-plus-storage task is thought to mimic skills used when listening to speech. The Gaps in Noise (GIN) test (Musiek 2004) could also be employed as a possible pre-fitting measure in the hearing aid selection process. The GIN measures the patient’s ability to process rapid temporal information, a cue that may be related to optimal compression speed (Gatehouse et al., 2006).

Figure 1 shows the relationship between aided SNR loss and results of the Reading Span Test as compression release time was varied for study participants. A statistically significant relationship was found between poor scores on the RST and performance with slow- vs. fast-acting compression. Specifically, our patients with good working memory performed better with fast-acting compression, while patients with poor working memory performed better with slow-acting compression—a result very similar to that found in laboratory studies (e.g., Rudner et al. 2009; 2011). However, in our study current generation hearing aids were used. Additionally, the aids were programmed in the clinic and features such as digital noise reduction and feedback reduction were activated. Because this real-world, clinical approach would be expected to introduce greater variability, it is encouraging to see the same pattern as shown in more controlled laboratory comparisons. While poor scores on the GIN test (i.e. poor temporal resolution) were associated with better performance with slow-acting than with fast-acting compression hearing aids, the relationship was not statistically significant. Our data suggest that the RST could be used during the pre-fitting appointment to identify patients who are more likely to benefit from slower compression release times.
Why use pre-fitting measures in the clinic?
Fig 1: Subjects with high cognitive scores tended to perform better with FAST-acting WDRC; subjects with poor cognitive scores performed better with SLOW-acting WDRC.

The hearing aid return for credit rate for participants of this study was 7% (compared to an industry average of nearly 16%; Strom 2008). Of course, this lower than average return for credit rate is probably not entirely due to use of the GIN and Reading Span tests during the pre-fitting appointment to identify compression parameters, as there are sure to be other differences across practices in patient population, audiologist expertise, etc. that may account for this lower than average return for credit rate. Nonetheless, spontaneous comments from the patients who were given these additional tests suggested high satisfaction and perceived value. Specifically, they perceived that their needs were being carefully and objectively considered, and reported high confidence that the extra testing provided would lead to the best outcome. Although a successful outcome cannot be guaranteed for every patient even when pretests are used, patients who approach the fitting process with a more positive outlook are more likely to be successful (Cox, Alexander & Gray, 1999).
Fig. 2: Subjects with poor gap detection ability tended to perform better with slow-acting WDRC, although the effect was not statistically significant. Subjects with good gap detection ability tended to perform similarly with both slow- and fast-acting WDRC

How do I find the time?
Assuming that the baseline audiological evaluation takes approximately 45 minutes, the GIN test (10-15 minutes, depending on implementation) can be added to the audiological evaluation. The hearing aid evaluation can be done in one concurrent or subsequent visit lasting 30-60 minutes, which includes the RST (10-20 minutes, depending on whether the long or short form is used). The additional information (here, temporal resolution and working memory) can be used to support the choice of signal processing or as a counseling mechanism to discuss individual differences with the patient. Dispensing and follow-up can be done as deemed appropriate by the audiologist. While more research is needed to determine which additional tests will be most useful, we support an approach which includes specific tests to guide selection of hearing-aid parameters. Figure 3 shows one possible protocol incorporating the RST and/or GIN test.
Fig. 3: Example of evaluation and fitting protocol which uses GIN or RST prefitting tests.

Final Thoughts
The study described here represents one possible approach to hearing aid selection. Dispensing only one sound-processing strategy is unlikely to be the best choice for all patients. Although more research is needed to determine what pretests should be used to select signal processing parameters, our work and others suggests those tests should include cognitive as well as auditory tests. With the marvelous technology available to us, the next step in our professions should be a customized, evidence-based approach to hearing aid selection.    
Convery, E., & Keidser, G. (2011). Transitioning hearing aid users with severe and profound loss to a new gain/frequency response: benefit, perception, and acceptance. J Am Acad Audiol, 22 (3), 168-80. Cox, R., Alexander, G., & Gray, G. (1999). Personality and the subjective assessment of hearing aids. J Am Acad Audiol. 10(1), 1-13.

Gatehouse, S., Naylor, G., & Elberling, C. (2006). Linear and nonlinear hearing aid fittings – 2. Patterns of candidature. Int J of Audiol, 45, 153-171.

Johnson, E.E. (2007). Survey explores how dispensers use and choose their preferred hearing aid brands, Hearing, 60 (3), 23-36.

Johnson, E.E. (2011). Determinants of Department of Veterans Affairs hearing aid brand dispensing by individual audiologist. J of Rehab Res and Dev, 48 (7), 865-880.

Johnson, E.E., Mueller, H.G., & Ricketts, T.A. (2009). Statistically derived factors of varied importance to audiologists when making a hearing aid brand preference decision. J Am Acad Audiol, 20 (1), 40-48.

Killion, M.C., Niquette, P.A., Gudmundsen, G.I., Revit, L.J., & Banerjee, S. (2004). Development of a quick speech-in-noise test for measuring signal-to-noise ratio loss in normal hearing and hearing impaired listeners. J Acoust Soc Am, 116 (4, pt. 1), 2395-405.

Lunner, T., & Sundewall-Thoren, E. (2007). Interactions between cognition, compression, and listening conditions: Effects on speech-in-noise performance in a two-channel hearing aid. J Am Acad Audiol, 18 (7), 604-617.

Musiek, F.E., & Shinn, J.B., Jirsa, R., Bamiou, D., Baran, J.A,m & Zaidan, E. (2004, December). GIN (Gaps-In-Noise) test performance in subjects with confirmed central auditory nervous system involvement. Ear and Hearing, 608-618.

Rudner, M., Foo, C., Jerker, R., & Lunner, T. (2009). Cognition and aided speech recognition in noise: Specific role for cognitive factors following nine-week experience with adjusted compression setting in hearing aids. Scand J of Psychol, 50, 405-418.

Rudner, M., Jerker, R., & Lunner, T. (2011). Working memory supports listening for noise for persons with hearing impairment. J Am Acad Audiol, 22 (3), 156-167.

Souza, P.E., & Sirow, L.W. (2012, March). Incorporating cognitive tests in the clinic. Poster presented at the American Academy of Audiology Annual Convention, Boston, MA.

Strom, K. (2007, May). Green lights and speed bumps: a review and regional sales analysis of the 2006-2007 hearing care market. Hearing Review, 25-28.

Taylor, B., & Mueller, H. G. (2011). Fitting and Dispensing Hearing Aids. San Diego, CA: Plural Publishing Inc.
We wish to thank Thomas Lunner for his assistance providing the Reading Span Test and for helpful comments concerning this study, and Cindy Bergman for her help with data collection. Work was partially supported by NIH (R01 DC012289) to the second author.
Lynn Sirow, Ph.D., Port Washington Hearing Center, Port Washington, NY and City University of New York. Email:

Pamela Souza, Ph.D., Communication Sciences and Disorders and Knowles Hearing Center, Northwestern University, Evanston, IL. Email: