P70Session 2 (Tuesday 13 January 2026, 14:10-16:40)Predicting speech in noise perception for cochlear implant listeners from a combination of measures of amplitude modulation rate discrimination and viability of the electrode-neural interface
Speech perception using a cochlear implant (CI) is reliant upon the listener’s ability to process envelope cues in multiple independent channels.
In this research we use two measures to assess these abilities: (1) the Amplitude Modulation Discrimination test for Cochlear Implants (AMCI) and (2) the Panoramic Electrical Compound Action Potential Method (PECAP). Boths measures aim to estimate neural health and current spread. The AMCI is a perceptual measure that assesses AM rate discrimination abilities (13 versus 40 Hz) at the cortical/sub-cortical level in the presence or absence of a speech-shaped noise masker on adjacent channels. The PECAP measures eCAPs using a forward-masking artefact-reduction technique for every combination of masker and probe electrodes at the most comfortable level, a peripheral measure. The children’s coordinate response measure (CCRM), an adaptive speech test using either a multi-talker or a speech-shaped noise masker was used to assess speech-in-noise perception.
A pilot study was conducted with ten adult cochlear implant listeners (Cochlear Nucleus device users) tested on five different target channels (19, 16, 13, 10, 7), to evaluate how the AMCI and PECAP measures relate to one another, to understand how they might relate to speech-in-noise perception and to determine the power calculation for a larger scale exploration of these effects.
The measures of current spread from the AMCI and PECAP seem to be related with a moderate-to-large effect size (r=0.60) but the measures of neural responsiveness had a smaller effect size (r=0.23) for the relationship, which was not significant for this pilot work. Different models were estimated for the prediction for speech-in-noise perception, with the best-fitting model including both the AMCI and PECAP measures as predictors (r=0.83; for both competing noise types).
The power analyses suggested that for some of the analyses the sample size was sufficient but to be able to fully understand the relationship between measures a larger sample of 26 participants are required.