evidence-based bimodal fitting bas van dijk · final evaluation of fitting flow bimodal benefit...
TRANSCRIPT
Evidence-based bimodal fitting
Bas van Dijk
Bimodal fitting
• Often done by separate
people.
• Difficult??
• What is the goal?
• What is needed is clear
guidance for clinicians,
based on objective data
that gives good, stable
outcomes.
Evidence-based fitting
Theresa Ching 2001
First fitting flow proposedDefault setting in HAPrograms matched
NAL-NL2 or Audiogram+
M1:Melbourne study (N=17)Speech understanding and preference
Modified proposal Matched signal processing
M2: Melbourne study (N=10)Preference only
New proposalMost parameters back to default
programs matched
M3:Melbourne study (N=17)Fine Tuning
Erasmus studyFine Tuning
M4: Melbourne study (N=17)Final evaluation of fitting flow
Bimodal benefit
NAL-NL2 better than own
fitting, No evidence we
need adjustments or
complex loudness
balancing procedure.
Smart Hearing Alliance
Melbourne trial (M1-4)
• 17 experienced bimodal recipients (N6 users, different HAs)
• All converted to Enzo2, Enzo29 or Lynx depending on HL
• Average age 69 years (range 52 to 83 years)
• Average period of implant use 5.3 years (range 0.8 to
11.5 years)
M2: Matched parameter proposal
• Adjusted parameters for best match
– Wind noise reduction
– NR strength
– Environmental optimizer
– Directionality
• Preference take-home study in 10 subjects (out of the original
17)
• 1 person preferred the new setting, 1 person no preference, the
rest clearly preferred the original setting.
• 3 people reported loudness inbalance with new setting, zero
with old setting.
• Reports of speech being unclear and HA not loud enough in
some noisy situations.
Theresa Ching 2001
First fitting flow proposedDefault setting in HAPrograms matched
NAL-NL2 or Audiogram+
M1:Melbourne study (N=17)Speech understanding and preference
Modified proposal Matched signal processing
M2: Melbourne study (N=10)Preference only
New proposalMost parameters back to default
programs matched
M3:Melbourne study (N=17)Fine Tuning
Erasmus studyFine Tuning
M4: Melbourne study (N=17)Final evaluation of fitting flow
Bimodal benefit
NAL-NL2 better than own
fitting, No evidence we
need adjustments or
complex loudness
balancing procedure.
Smart Hearing Alliance
Not preferred
M3+M4: Device Fitting
• Hearing Aid fitted to NAL-NL2 prescription at 50, 65
and 80 dB SPL
• Sound Processor CP900 (N6)
• SCAN with clinical defaults
M3: Method: Comparison of Loudness Balancing
Approaches
Two approaches evaluated
1. Female continuous discourse at 65 dB SPL
2. International Speech Test Signal (ISTS)
Presented at multiple levels of 55, 65 and 75 dB SPL
Filtered into low-frequency (< 500 Hz), mid-frequency (≥ 500 Hz and ≤ 1000 Hz) and high-frequency (> 1000 Hz)
components
Hearing aid gain adjustments were applied until subject judged the sound from hearing aid and cochlear implant sound processor was centred in the head when listening to both together.
M3-Method: Comparison of Loudness Balancing
Approaches
Preferred gain measures:
Female Continuous Discourse at 65 dB SPL
• Gain adjustments applied based on 65 dB SPL signal
only
International Speech Test Signal (ISTS)
• Gain was adjusted in response to subject reports for
each frequency band and level
• Where there were no thresholds ≤ 90 dB HL for a
particular frequency band, that frequency range was
not used for the balancing.
For each approach, the resulting REIG was measured.
M3-Method: Comparison of Loudness Balancing
Approaches
Bimodal performance measures:
• Speech perception, SRT at 65 dB SPL with adaptive
4TB, S0N0
• Localisation, 13 speaker array in 180 degree
configuration in the horizontal plane, stimuli Pink
noise bursts. The maximum presentation level was
68 dB SPL and level was randomly jittered by up to
8dB in steps of 1 dB
M3-Results: Comparison of Loudness Balancing
Approaches
Average preferred gain preferences at 65 dB SPL (n=17)
Continuous discourse – 15/17 within ±5dB NAL-NL2
ISTS Multi Level – 14/17 within ± 5 dB NAL-NL2
Continuous
Discourse ISTS Multi
Level
M3-Results: Comparison of Loudness Balancing
Approaches
Preferred 3FA gain
Significant correlation between 65
Discourse and 65 ISTS Multi
conditions
NO correlation between corrections
(suggesting that on average one
could just fit to NAL-NL2 without
loudness balancing).
M3:Results: Comparison of Loudness Balancing
Approaches
Bimodal performance
SRT – no significant difference between approaches (n=14)
Localisation - no significant difference between approaches (n=15)
Better
Performance
Theresa Ching 2001
First fitting flow proposedDefault setting in HAPrograms matched
NAL-NL2 or Audiogram+
M1:Melbourne study (N=17)Speech understanding and preference
Modified proposal Matched signal processing
M2: Melbourne study (N=10)Preference only
New proposalMost parameters back to default
programs matched
M3:Melbourne study (N=17)Fine Tuning
Erasmus studyFine Tuning
M4: Melbourne study (N=17)Final evaluation of fitting flow
Bimodal benefit
NAL-NL2 better than own
fitting, No evidence we
need adjustments or
complex loudness
balancing procedure.
Simple LB as good as complex.
Maybe not needed at all?
Smart Hearing Alliance
Not preferred
SHA- Bimodal Fitting Workflow
NAL-NL2 prescription
• Shown to provide suitable frequency response and gain characteristics for bimodal listeners (English et al, 2016)
Hearing Aid Program Configuration
• Best match to CP900 defaults of SCAN & Custom
• P1 Soft Switching (directionality adjusted based on listening environment)
• P2 Omni-Directional
Simple Loudness Balancing
• Listening with two ears
• Adjust gain in hearing aid based on loudness judgement using continuous discourse presented at conversational speech level
Recommended Workflow:
M4-Method: Acceptance of Bimodal Fitting Workflow
• Each subject was provided with a minimum of two weeks take home experience with the Linx2 9 or Enzo2 9 hearing aid and the CP900 sound processor fitted using the recommended workflow
• Overall satisfaction level obtained comparing the hearing aid fitting (as provided using the loaner hearing aid and workflow) and their own hearing aid fitting at the time of enrolment
• Subjective feedback collected to explore the potential fine-tuning refinements that might improve satisfaction from that reported from direct application of the bimodal fitting workflow
M4-Results: Acceptance of Bimodal Fitting Workflow
• 16/17 subjects were satisfied with the
bimodal fitting workflow after take-home
use, which included individual fine-
tuning.
• 12 of the subjects reported that they
were ‘very satisfied’ with the fitting, and
4 reported being ‘somewhat satisfied’.
• 1 subject reported being neither
satisfied nor dissatisfied with the fitting.
M4-Results: Acceptance of Bimodal Fitting Workflow
Analysis of the fine-tuning adjustments indicated that:
• Three subjects reported preference for the hearing aid to be louder than the sound processor, ie. bimodal balance was notpreferred
• Two subjects had a frequency response in their own hearing aid that differed substantially from NAL-NL2. They reported being ‘somewhat satisfied’ with the workflow, may have benefited from further fine-tuning to match their own hearing aids more closely.
• One subject who reported being neither satisfied or dissatisfied had experienced feedback problems. Was unavailable for further appointments to trial a more occluding mould.
Theresa Ching 2001
First fitting flow proposedDefault setting in HAPrograms matched
NAL-NL2 or Audiogram+
M1:Melbourne study (N=17)Speech understanding and preference
Modified proposal Matched signal processing
M2: Melbourne study (N=10)Preference only
New proposalMost parameters back to default
programs matched
M3:Melbourne study (N=17)Fine Tuning
Erasmus studyFine Tuning
M4: Melbourne study (N=17)Final evaluation of fitting flow
Bimodal benefit
NAL-NL2 better than own
fitting, No evidence we
need adjustments or
complex loudness
balancing procedure.
Simple LB as good as complex.
Maybe not needed at all?
Flow well accepted and easy to use
Balancing not preferred by all
Smart Hearing Alliance
Not preferred
Conclusions (M1,M2,M3,M4, Erasmus)
• NAL-NL2 gives suitable amplification (M1,M4, Erasmus)
• Loudness balancing only gives small correction (M3,
Erasmus)
• Complex loudness balancing (frequency specific and/or
multi-level) no benefit over simple (M3,Erasmus)
• No difference in speech understanding or localization
between the approaches (M3, Erasmus)
• No direct evidence that we need loudness balancing at all
(M3, Erasmus)
• The majority subjects were satisfied with the fitting
generated using the bimodal fitting workflow. (M4)
Acknowledgements
• Resound: Astrid Haastrup, Chantal Crins, Tammy Stender
• Erasmus: Jantien Vroegop, Andre Goedegebure
• Cochlear: Ruth English, Kerrie Plant, Birgit Philips
• Publications:
– English et al. 2016, International Journal of Audiology
– Vroegop et al. 2016, submitted Audiology&Neurotology