Neural Prediction to Enhance Language Outcomes in Children with Cochlear Implant

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT Although cochlear implantation (CI) is the most effective method for managing severe to profound sensorineural hearing loss, children with CI as a group perform at about 15th percentile of their normal-hearing peers on language measures. Most intriguingly, their language outcomes are highly variable at the individual level, despite implantation at a young age. Using pre-surgical brain magnetic resonance imaging (MRI) scans, done as part of the routine clinical evaluation, as well as AI-enabled analytical methods, our research will construct neural predictive models to forecast individual-level language outcomes in English- and Spanish-learning children up to 4 years after surgery. The clinical utility of these models will also be evaluated by investigating the extent to which the models' prediction is associated with the degree to which a child responds to a program of intensive communication treatment. We hypothesize that during the 4 years immediately after surgery, young children with CI follow three stages of language development: 1) global attention to spoken language as the child acclimatizes to electric auditory input about language, 2) encoding of phonological patterns with sufficient information to develop auditory-based lexical representations, and 3) development of spoken language syntax to communicate orally in longer utterances. We also hypothesize that the integrity of brain networks associated with higher-order cognitive, auditory and syntactic processing differentially contributes to language outcomes across these three stages in monolingual English-learning children with CI (Aim 1). We further hypothesize that these networks are largely invariant for typologically similar languages during the first stage, but that the contribution of the auditory network would be prolonged and require higher-order cognitive networks to an even greater extent to language outcomes for Spanish-English bilingual children with CI (Aim 2). As part of our current R21 project, we have developed a standardized clinical evaluation and follow-up protocol across different CI centers that will facilitate the investigations required to achieve Aims 1 and 2. Aim 3 concerns the interaction between neural prediction of outcomes and behavioral treatment. The main CI center of this project will enroll monolingual English-learning children for an intensive, Parent-Implemented Communication Treatment (PICT) program, which is the only treatment program to date whose effectiveness has been supported by a randomized controlled trial. We will evaluate whether neural prediction of language outcomes is inversely related to the degree of language gains from PICT. Specifically, we hypothesize that the more severe the predicted language impairment based on our neural predictive algorithms, the more the child could benefit from PICT. Our translational research program will advance the field of communication disorders in technological, theoretical and clinical innovations. It will be among the first to demonstrate that a predict-to- prescribe approach to holistically treat hearing loss is feasible, cost-effective and can lead to optimization of language outcomes of all children with CI.