Early Spectral Resolution Predicts Later Speech Recognition in Adult Cochlear Implant Recipients

To investigate the trajectory of spectral resolution in adult cochlear implant (CI) recipients and determine whether early spectral resolution measures can predict later speech recognition outcomes.

Prospective, longitudinal study of 531 postlingually deafened adult CI recipients. Spectral resolution (EasyQSMD) and speech recognition (CNC words, AzBio sentences in quiet and noise) were evaluated from activation through 1-year post-activation. Growth curve models were fit to EasyQSMD thresholds across all timepoints to generate mode-based estimates of early spectral resolution at 1-month post-activation. These model-derived early thresholds were then used to predict later speech recognition outcomes (best of 6-months and 1-year) through linear regression analyses.

EasyQSMD thresholds improved from activation to 1-week post-activation, then stabilized through the first year. Model-derived early EasyQSMD thresholds significantly predicted later speech recognition outcomes, explaining 3%-8% of the variance in performance.

Spectral resolution rapidly stabilizes, reaching asymptotic performance by 1-week post-activation. The EasyQSMD provides a non-linguistic assessment tool that demonstrates significant predictive value for subsequent speech recognition outcomes. This early predictive capability enables the identification of poor performers during a critical window when speech recognition skills are still developing and the brain's neuroplasticity is at peak potential. Integrating spectral resolution assessment into early post-activation clinical protocols could improve individualized realistic expectations counseling and guide targeted interventions-potentially leading to improved overall CI adoption rates. The brief administration time and independence from linguistic content make the EasyQSMD particularly valuable for multilingual populations and during the earliest stages of auditory rehabilitation post-CI.