A Computational Approach to Optimal Deactivation of Cochlear Implant Electrodes

Primary outcomes

1. Change in Vowel and consonant identification

   Time frame: Baseline, Month 6

   Vowel and consonant identification testing will include 9 vowels in /hVd/ context and 16 consonants in /aCa/ context presented in quiet. Each phoneme will be presented 15 times. Vowel and consonant testing will be done during initial speech testing using clinical CI settings. Subject confusion matrices (i.e. number of correct responses and specific confusions) will serve as the basis for model construction.

2. Change in Word recognition

   Time frame: Baseline, Month 6

   Word recognition testing will consist of a Consonant-Nucleus-Consonant (CNC) monosyllabic word test that includes 30 lists of 50 words each presented in quiet. Participants will complete three CNC word lists per testing session, and word lists will be blocked across testing sessions so no word list is repeated. Word recognition scores will be used to evaluate performance with clinical and experimental CI settings. Word scores with clinical settings will also be used for model development.

3. Change in Sentence recognition

   Time frame: Baseline, Month 6

   Sentence recognition testing will consist of AzBio sentences tested in +10 decibel (dB) signal-to-noise ratio (SNR) multi-talker babble. The test includes 33 lists of 20 sentences with approximately 150 words per list. Participants will complete two lists per testing session, and lists will be blocked across testing sessions so no lists are repeated. Sentence recognition scores in noise will help evaluate generalization of performance with experimental settings to other speech tasks and in other listening conditions. Speech stimuli will be presented at 65 dB sound pressure level (SPL) (C-weighted) in a sound attenuated room as measured from CI user's microphone.

4. Change in Electrode discrimination

   Time frame: Baseline, Month 6

   Electrode discrimination will be measured as a JND relative to every active CI electrode in both basal and apical directions. Individual electrodes will be stimulated with 500ms pulse trains at comfortable loudness-balanced levels. JND for each reference electrode will be measured adaptively using a 4-interval 2-alternative forced-choice (2AFC) 2-down/1-up procedure. Electrode discrimination Just Noticeable Differences (JNDs) will serve as inputs to build the proposed subject-specific models.

5. Change in Pitch Scaling

   Time frame: Baseline, Month 6

   Subjects will be instructed to assign a number from 1 to 100 that corresponds to their perceived pitch of a stimulated electrode (100 highest in pitch). Each electrode will be stimulated once per block using 500ms pulse trains at comfortable loudness-balanced levels, presented in random order. A trimmed mean of pitch ratings (discarding lowest and highest ratings) for the last eight of ten blocks will be used to determine the final score. Pitch scaling will be used to evaluate whether electrodes are perceived as well ordered in pitch, and will be measured with clinical CI settings.

6. Change in Sound Quality

   Time frame: Baseline, Month 6

   A shortened version of the speech, spatial, and qualities of hearing scale (SSQ12) will be administered when using clinical CI settings, and after using experimental CI settings for 1.5 months. The SSQ12 is a self-assessment questionnaire measuring hearing disability across three domains. The shortened version, SSQ12, includes 12 questions across the three domains. Each question is answered using a 10-point scale ranging for poorest (i.e. 0) to best (i.e 10) performance. The average score across questions will be used to evaluate sound quality across CI settings conditions and across subjects. The SSQ12 has previously been validated in CI users, and is highly correlated with results from the original 49 item SSQ.