Support Fitting Frequency Compression Hearing Aids: A review of current evidence Presenter: Collaborator: Susan Scollie Danielle Glista This work was supported by: Canadian Institutes of Health Research, National Science and Engineering Research Council, Masons Help-2-Hear Foundation and Phonak AG Special thanks to: Richard Seewald, Vijay Parsa, Melissa Polonenko, Marlene Bagatto, and Jacob Sulkers for their contributions Some Types of evidence to consider: Studies of bandwidth: Does it matter if we provide audibility of high frequency speech cues to children who use hearing aids? Are we providing these cues? Electroacoustic information: What is the signal processor doing to speech? Can we measure that? Behavioural studies of outcome: Is there benefit when we apply this processing strategy? Studies of bandwidth: Recent studies (many by Stelmachowicz, Pittman, Moeller): Children who have hearing loss need greater access to the high frequency cues of speech to recognize and learn speech sounds as well as their normally hearing peers. High frequency speech sounds appear to be delayed in the early phonemes produced by children who use hearing aids. The authors speculated that the bandwidth of typical hearing aids may not reach the spectral energy present in many fricatives. Studies of bandwidth: Stelmachowicz et al 24 Female & child s versus male s Studies of bandwidth: Stelmachowicz et al 21 NH listeners: Hz seems adequate. HI listeners: more than a Hz bandwidth was needed in this study. 1
Studies of bandwidth: Pittman 28 Faster word learning with extended bandwidths. Hz versus 9 Hz, using lab equipment. Clinical rationale for frequency lowering: Bandwidth matters: Children need access to the high frequency sounds of speech, to understand and monitor: See: Moeller et al, 27, a review article by Stelmachowicz et al (24), Pittman 28, and various other studies I`ve only mentioned a few here. But: If audibility cannot be providedvia the available bandwidth and gain/output, is it beneficial to lower the cues to an audible frequency range? Bandwidth limitations for severe losses: Frequency Lowering (FL): two types Hearing aid A severe sloping loss. responses and targets for speech (input at 65 db SPL) Hearing Aids A and B are both modern devices, and are at maximum settings in this region. Both fall below threshold above 2 Hz (speech peaks are audible to about 25 Hz not shown). Frequency Compression (FC) E.g.: Phonak SoundRecover Frequency Transposition (FT) E.g.: Widex AudibilityExtender Review: Simpson (29), Trends in Amplification Below cutoff Below cutoff Above cutoff Above cutoff Spectral analysis of FC: Spectrographic analysis of FC: f3 - original - no NFC f3 - CR = 2, CF = 4k 5 5 1 2 3 4 Time (s) 1 2 3 4 Time (s) f3 - CR = 2, CF = 3k f3 - CR = 2, CF = 2k 5 5 1 2 3 4 Time (s) 1 2 3 4 Time (s) 2
What does FC sound like? 3kHz, 2:1 and 6:1 2kHz, 2:1 Original Fitting Method (pediatric): 1) Provide more audibility of high frequency cues than is possible with a well-fitted device. The frequency response is based on DSL5 child to maximize the bandwidth of the fitting without frequency lowering. 2) We verify using measuresthat show us audibility of specific high frequency speech bands (see Glista & Scollie, AudiologyOnline 29) Clinical verification Sample fitting: Running speech can be used. Broadband noises can be used (not shown). Speech in the lower band (unchanged). Speech in the upper band (compressed). Audibility is better here (judging by peak SL of at least 1 db). Audibility is limited here Sample fitting: Sample fitting (with FC): SSSSS SSSSS SHHHH SHHHH 3
Live /s/ versus calibrated /si/ at 7 db SPL: shhh (FC on) sss (FC on) shi (FC on) si (FC on) Clinical verification Live speech productions of /s/ and /sh/ can be used. They are not calibrated, but provide an informal way to see change in the frequency location of speech sounds. They provide a realistic view of the bandwidth and frequency location of the frication bands of /s/ and /sh/. SSSSS SSSSS SHHHH SHHHH Without FC With FC New clinical tests A new test signal from the Verifit allows us to see if a high frequency band is lowered. Run with & without SoundRecover. Test at 3.1, 4k, 5k or 6.3kHz. Mayoffer a calibrated alternative to live voice /s/ and /sh/. Narrower bandwidth though. Level - Level - Level - Original signal (speech passages, used for regular verification) Frequency - High frequency speech band Low frequency Notch speech energy (empty area) Frequency - Lowered speech band Low frequency Notch speech energy (empty area) Verifit Speech Bands with/without: This speech signal has been notch filtered With FC, here. 63 Hz above A high frequency threshold: band is left (63 Hz in this example.) The notch lets you observe lowering of the high frequency band. Without FC, the 63 Hz band is below threshold. Frequency - Our uses of this new test Is the 6.3kHz region audible with vs. without SoundRecover? With Without Our uses of this new test With SoundRecover on, are the 4kHz and 6kHz regions overlapping? Our previous research tells us that overlapping /s/ and /sh/ using high SoundRecover settings led to rejection. The 4kHz and 6.3kHz bands are similar to /s/ and /sh/. Little overlap 4k and 6.3 khz A lot of overlap This setting may be good to try. This setting may be using too much. 4
Some studies of frequency lowering: (other studies have been done in adults, plus other non-peer reviewed in kids) FC in children: Glista et al., 29a FC improves detection/recognition (group vs individual) Significant candidacy factors (hearing loss, age group) FT (various types) in children: Miller-Hansen et al, 23, MacArdle et al, 21 FT improves detection & recognition (group vs individual) Auriemmo et al, 29 FT + training improved consonant recognition (other outcomes) Smith et al, 29 FT improved consonant recognition Efficacy? Experimental design factors Baseline FC should be compared to the best possible fitting. Does this change over time as the fittable bandwidth extends? Candidacy? Aid FC is best evaluated within-devices. Allows us to hold all other device variables constant. Fitting FC settings should be appropriate to the individual. Optimal settings are not yet known, but fitting, tuning, and verification are possible. Time An acclimatization period may be necessary. What does this mean for studies comparing FL strategies? Measures As with all hearing aid research, blinding is needed for subjective measures. Sensitive tests are needed but may not test all speech sounds a test battery? On to the field trial Study design. 24 patients: 11 children and 13 adults A wide range of hearing losses from mild through profound. Glista, D., Scollie, S., Bagatto, M. Seewald, R., Parsa, V., Johnson. A. (accepted). Evaluation of nonlinear frequency compression: Clinical outcomes. International Journal of Audiology. Time course Objective Duration Participant intake Acclimatization phase NFC phase Multimemory phase Withdrawal testing Audiometric evaluation. Hearing aid fitting (CP). Range: 2 weeks to 3 Real world trial with CP. Practice tests. Real world trial with NFC. Outcome evaluation with NFC. Real world trial with user selectable NFC. Evaluation of real world preferences. Outcome evaluation without NFC. months Mean: 4.17 weeks Range: 3 weeks to 1.3 years Mean: 1.75 weeks Range: 2 weeks to 5 months Mean: 5.58 weeks Fitting recap: Hearing aids were first fitted using DSL, without NFC. Then we individualized it, and activated it, testing for: Improvements in audibility /s/-/sh/ separation in frequency Sound quality User reaction (if possible) Results after several weeks of use Outcomes battery: Aided detection thresholds of the phonemes s, ʃ Recognition of high frequency consonants: /ʧ, d, f, ʤ, k, s, ʃ, t, d, z/ spoken by two female talkers Identification of word-final pluralson 15 words: ant, balloon, book, butterfly, crab, crayon, cup, dog, fly, flower, frog, pig, skunk, sock and shoe Double blind subjective preference. Speech production. 5
Speech sound detection improved. Consonant & word-final plural recognition improved. Some adults benefit more. Children s results look better. Audiometric & age candidacy. Summary of outcomes Significant predictors of outcome: Age group (adult versus child) Better ear high frequency pure tone average The lowest frequency at which the audiogram had a severe loss (drop off frequency) db HL db HL Frequency -2 2 12 Frequency -2 2 12 On average, the NFC processor improved speech sound detection thresholds, as well as consonant and plural recognition scores; vowel perception was not significantly changed. Individual results indicated that age group and degree and configuration of hearing loss were related to NFC benefit and to preference. Variance in individual outcome results was considerable. Individual determination of candidacy is warranted when considering NFC use in clinical application. 6
Does it affect speech production? Trial participants were recorded while repeating sentences: Give me your socks. She got a shot. Soup is good food. My shoes are new. I see the fox. And while answering questions (e.g., Tell me how you would plant a seed.). Before NFC and after 9.5 weeks mean acclimatization Range: 6 to 14 weeks Electroacoustic and subjective analyses Long term average speech spectra (all children). 1/3 Octave Band Level (db SPL) 75 7 65 55 5 45 Normal Hearing Children Hearing Impaired Children (Conv) Hearing Impaired Children (FC) Vowel space (one child): Can you tell me how to plant a seed? F1 (Hz) 9 7 5 Conv Area = 65412.736 /i/ F2 - /u/ F2 = 31.6774 /a/ /i/ /u/ F1 (Hz) 9 7 5 FC Area = 148668.5779 /i/ F2 - /u/ F2 = 736.4264 /a/ /i/ /u/ db HL Frequency -2 2 2, 6:1 Significant benefit with NFC Now I hear /s/ where before there was only silence Difficulty adjusting 12 3 15 2 25 3 F2 (Hz) 3 15 2 25 3 F2 (Hz) Can you tell me how to plant a seed? and how to make a sandwich? db HL -2 2 12 Frequency 35 Hz, 2:1 No objective benefit because at ceiling on all measures at baseline. Significant blinded preference Follow up The FC processor was included in a commercial product & labelled SoundRecover. Our participants, who by now had been wearing prototypes for some time, were called back in to have their aids replaced with commercial devices & retesting. 7
Prototype versus commercial: The commercial aid had more gain and a slightly different frequency response. The frequency compression parameters are not as adjustable, and limit the CR to 4:1. We matched as closely as possible. Participants were retested with prototypes so that commercial/prototype differences shown are not attributable to acclimatization. Paper Three: Comparison of prototype to Naida, as described in Glista et al, (29b). Glista, D., Scollie, S., Polonenko, M., & Sulkers, J. (29, November). Prototype nonlinear frequency compression versus SoundRecover : A comparison of performance in children. The Hearing Review. Participants: 1 child participants from the Glista et al. (29a) study Devices: Savia 311 and 411, modified to include prototype NFC Naida V SP and UP with SoundRecover NFC Fitting: Participants wore Savia hearing aids for at least one year (M = 2.3) prior to being refit with Naida Frequency response, amplitude compression, frequency compression and other signal processing features of Naida were matched to each participant s Savia Glista, D., Scollie, S., Polonenko, M., & Sulkers, J. (29, November). Prototype nonlinear frequency compression versus SoundRecover : A comparison of performance in children. The Hearing Review. Devices were well matched. Matching NFC Cut-Off and Ratio Ear Canal Level (db SPL) 12 (a) Right Ear Fit to Targets Targets Naida Measured REAR Prototype Measured REAR Threshold 2 Ear Canal Level (db SPL) 12 (b) Left Ear Fit to Targets Targets Naida Measured REAR Prototype Measured REAR Threshold 2 Subject Savia cut-off (khz) Savia compression ratio values Naida cut-off (khz) Naida compression ratio values 232 3 2:1 32 3.1:1 234 19 8:1 1 4:1 2 2 6:1 2 3.6:1 261 35 2:1 35 2.1:1 262 R 3 L 27 R 4:1 L 6:1 29 4:1 263 23 6:1 22 4:1 265 15 4:1 15 4:1 266 1 7:1 1 4:1 268 32 3:1 29 2.7:1 269 3 3:1 3 3.7:1 8
How to read the next few slides: How to read the next few slides: Vowel Recognition score (perecnt correct) 9 7 5 Across these time points. These 2 test points are from the original trial These 2 test points are from about a year later Plural Recognition score (percent correct) 9 7 5 If scores look like this Benefit Vowel Recognition score (perecnt correct) 9 7 5 Across these time points. These 2 test points are from the original trial These 2 test points are from about a year later Plural Recognition score (percent correct) 9 7 5 If scores look like this No change 3 3 3 3 s, ʃdetection same or better with FC; some cases better with Naida. Consonant Recognition: some children benefit from NFC. Detection Threshold (db HL) -2 2 /s/ Detection Threshold (db HL) -2 2 / / Consonant Recognition score (percent correct) 9 7 5 3 Plural Recognition: most children benefit from FC. Vowel Recognition largely unaffected by NFC. Plural Recognition score (percent correct) 9 7 5 Vowel Recognition score (perecnt correct) 9 7 5 3 3 9
But what about individuals? Some children showed a lot of change over time. In the previous slides, the same children are marked with the same colours throughout. Let`s look back at a few. Are we seeing possible acclimatization effects? Do children need an acclimatization period? Danielle Glista, Ph.D. Child Amplification Laboratory National Centre for Audiology, University of Western Ontario What is auditory acclimatization? New acoustic information Example: new audibility of speech cues post hearing aid fitting Time to acclimatize Systematic change in auditory performance From the Eriksholm workshop on Auditory Deprivation and Acclimatization (Arlinger et al., 1996) Why study auditory acclimatization? Work by Stuart Gatehouse and the Eriksholm Workshop on Auditory Deprivation and Acclimatization (1995): Auditory acclimatization is a real phenomenon with importance research/clinical implications Evidence suggests the mean reported improvement in benefit over times ranges from to 1% (across speech materials and presentation conditions) (Arlinger et al., 1996) Why study auditory acclimatization? Study design Previous research on nonlinear frequency compression (FC) and speech perception benefit suggests: Considerable performance variability at the level of the individual - adult and child data (Simpson, 29) Pediatric pilot data provides informal evidence of an acclimatization effect for some listeners (Glista et al, 29) As with all fittings involving new, complex signal processing, adaptation time becomes important Baseline Phase (No FC) Treatment Phase (with FC) Withdrawal Phase (No FC) Structure/Objective Real-world usage DSL v5. with adjustments to preference 2-3 testing sessions Stopping criterion: Asymptotic performance Goal: Minimize practice effects and/or acclimatization effects from previous fitting 4 testing sessions,spaced 2 weeks apart+ 2 monthly testing sessions Goal: Track time course/magnitude of an acclimatization effect 1 testing session FC disabled in lab only Goal: Establish FC effect post-acclimatization 1
Case Study 11 years of age Exposure to ototoxic medication Long-term, fulltime HA user Suspected dead regions (TEN test: Moore, Glasberg & Stone, 24) Hearing Threshold Level (db HL) 2 12 CNT DR DR DR Fitting details Study worn aids = Naida IX SP, SoundRecover setting = 1 Hz cut-off, 4:1 ratio DSL v5., FC setting individualized (refer to AudiologyOnline: Glista & Scollie, 29) Is this enough? 63 Hz with FC Hz with FC Live / / is Hz without FC broader in bandwidth 63 Hz without FC Filtered high-frequency speech bands Screen captures from the Audioscan Verifit / / /s/ Live speech with FC enabled Results Speech recognition: Plurality UWO Plurals Test Results Speech recognition: Consonants DFD Test (Cheesman & Jamieson, 1996) Crayon Crayons Baseline Treatment Withdrawal Improvement atil Time course Results Discrimination of /s/ vs. / / see she see Results Detection of /s/ and / / Ling-6 Sound Test mmm ooo aaa sss eee shh 11
Clinical implications Overall, significant speech perception benefit was reported with FC compared to without FC Acclimatization trends with FC: Benefit change ranged from to 17%, across measures Significant acclimatization trends were observed after approximately 6 weeks or longer Two unique acclimatization patterns where exemplified: Gradual improvement over time Improvement after a specific period of acclimatization (S-shaped curve) Further cases are currently under analysis Clinical implications Speech recognition tests at the time of fitting may not reflect later performance. Testing before and after an acclimatization period may reveal changes over time. Selected References: Thank you for listening Boretzki, M., Kegel, A. 29. The benefits of nonlinear frequency compression for people with mild hearing loss. Audiology Online, November. Glista D, Scollie S. 29. Modified verification approaches for frequency lowering devices. Audiology Online, November. Glista, D., Scollie, S., Bagatto, M., Seewald, R., Parsa, V., Johnson, A. 29a. Evaluation of nonlinear frequency compression: Clinical outcomes. International Journal of Audiology, Vol. 48, No. 9, Pages 632-644. Glista, D., Scollie, S., Polonenko, M. and Sulkers, J. 29b. A Comparison of Performance in Children with Nonlinear Frequency Compression Systems. Hearing Review, November. MacArdle, B. M., West, C., Bradley, J., Worth, S., Mackenzie, J., and Bellman, S. C. 21. A study of the application of a frequency transposition hearing system in children. British Journal of Audiology 35: 17-29. Miller-Hansen, D. R., Nelson, P. B., Widen, J. E., and Simon, S. D. 23. Evaluating the benefit of speech recoding hearing aids in children. American Journal of Audiology 12(2): 16-113. Moeller, M. P., Hoover, B., Putman, C., Arbataitis, K., Bohnenkamp, G., Peterson, B., et al. 27. Vocalizations of infants with hearing loss compared with infants with normal hearing: Part I--phonetic development. Ear and Hearing, 28(5), 5-627. Pittman, A. 28. Short-Term Word-Learning Rate in Children With Normal Hearing and Children With Hearing Loss in Limited and Extended High-Frequency Bandwidths. Journal of Speech, Language, and Hearing Research Vol.51 785-797. Stelmachowicz, P. G., Pittman, A. L., Hoover, B. M., Lewis, D. E., and Moeller, M. P. 24. The importance of highfrequency audibility in the speech and language development of children with hearing loss. Archives of Otolaryngology - Head and Neck Surgery 13(5): 556-562. Wolfe, J., John, A., Schafer, E., and Caraway, T. (in press). Evaluation of non-linear frequency compression for children with moderate hearing loss. Journal of the American Academy of Audiology. 12