Clinic-based speech recognition measures guide cochlear implant (CI) candidacy decisions but often fail to capture the full range of CI users’ outcomes and experiences—particularly those of poor performers underrepresented in research. To address these limitations, we examined individual performance differences among 40 CI users who completed virtual speech recognition and talker discrimination tasks. The tasks varied in terms of speaker characteristics, semantic context, and lexical content. Participants were classified as high or low performers based on their performance across tasks. We hypothesized a cascade effect where high performers would score consistently well and low performers consistently poorly rather than task-specific weaknesses. Strong correlations were observed among the speech recognition results. The resulting group profiles showed significant differences in performance for most tasks which supported our hypothesis. Density plots and classification trees further highlighted the tasks that better separated the performance groups. Speech recognition tasks demanding greater neurocognitive engagement were more predictive of performance group, but talker discrimination tasks can still offer a quick and inclusive way to identify underperforming individuals. Collectively, these results illustrate how remote, multidimensional assessments can enhance the sensitivity and inclusivity of CI outcome measurements while also facilitating the design of more ecologically valid measures.
Street et al. (Wed,) studied this question.