Dataset Information

A semi-supervised Support Vector Machine model for predicting the language outcomes following cochlear implantation based on pre-implant brain fMRI imaging.

ABSTRACT:

Introduction

We developed a machine learning model to predict whether or not a cochlear implant (CI) candidate will develop effective language skills within 2 years after the CI surgery by using the pre-implant brain fMRI data from the candidate.

Methods

The language performance was measured 2 years after the CI surgery by the Clinical Evaluation of Language Fundamentals-Preschool, Second Edition (CELF-P2). Based on the CELF-P2 scores, the CI recipients were designated as either effective or ineffective CI users. For feature extraction from the fMRI data, we constructed contrast maps using the general linear model, and then utilized the Bag-of-Words (BoW) approach that we previously published to convert the contrast maps into feature vectors. We trained both supervised models and semi-supervised models to classify CI users as effective or ineffective.

Results

Compared with the conventional feature extraction approach, which used each single voxel as a feature, our BoW approach gave rise to much better performance for the classification of effective versus ineffective CI users. The semi-supervised model with the feature set extracted by the BoW approach from the contrast of speech versus silence achieved a leave-one-out cross-validation AUC as high as 0.97. Recursive feature elimination unexpectedly revealed that two features were sufficient to provide highly accurate classification of effective versus ineffective CI users based on our current dataset.

Conclusion

We have validated the hypothesis that pre-implant cortical activation patterns revealed by fMRI during infancy correlate with language performance 2 years after cochlear implantation. The two brain regions highlighted by our classifier are potential biomarkers for the prediction of CI outcomes. Our study also demonstrated the superiority of the semi-supervised model over the supervised model. It is always worthwhile to try a semi-supervised model when unlabeled data are available.

SUBMITTER: Tan L

PROVIDER: S-EPMC4714644 | biostudies-literature | 2015 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A semi-supervised Support Vector Machine model for predicting the language outcomes following cochlear implantation based on pre-implant brain fMRI imaging.

Tan Lirong L Holland Scott K SK Deshpande Aniruddha K AK Chen Ye Y Choo Daniel I DI Lu Long J LJ

Brain and behavior 20151012 12

<h4>Introduction</h4>We developed a machine learning model to predict whether or not a cochlear implant (CI) candidate will develop effective language skills within 2 years after the CI surgery by using the pre-implant brain fMRI data from the candidate.<h4>Methods</h4>The language performance was measured 2 years after the CI surgery by the Clinical Evaluation of Language Fundamentals-Preschool, Second Edition (CELF-P2). Based on the CELF-P2 scores, the CI recipients were designated as either e ...[more]

PMID: 26807332

Similar Datasets

Project description:Cochlear implantation is a surgical alternative to traditional amplification (hearing aids) that can facilitate spoken language development in young children with severe to profound sensorineural hearing loss (SNHL).To prospectively assess spoken language acquisition following cochlear implantation in young children.Prospective, longitudinal, and multidimensional assessment of spoken language development over a 3-year period in children who underwent cochlear implantation before 5 years of age (n = 188) from 6 US centers and hearing children of similar ages (n = 97) from 2 preschools recruited between November 2002 and December 2004. Follow-up completed between November 2005 and May 2008.Performance on measures of spoken language comprehension and expression (Reynell Developmental Language Scales).Children undergoing cochlear implantation showed greater improvement in spoken language performance (10.4; 95% confidence interval [CI], 9.6-11.2 points per year in comprehension; 8.4; 95% CI, 7.8-9.0 in expression) than would be predicted by their preimplantation baseline scores (5.4; 95% CI, 4.1-6.7, comprehension; 5.8; 95% CI, 4.6-7.0, expression), although mean scores were not restored to age-appropriate levels after 3 years. Younger age at cochlear implantation was associated with significantly steeper rate increases in comprehension (1.1; 95% CI, 0.5-1.7 points per year younger) and expression (1.0; 95% CI, 0.6-1.5 points per year younger). Similarly, each 1-year shorter history of hearing deficit was associated with steeper rate increases in comprehension (0.8; 95% CI, 0.2-1.2 points per year shorter) and expression (0.6; 95% CI, 0.2-1.0 points per year shorter). In multivariable analyses, greater residual hearing prior to cochlear implantation, higher ratings of parent-child interactions, and higher socioeconomic status were associated with greater rates of improvement in comprehension and expression.The use of cochlear implants in young children was associated with better spoken language learning than would be predicted from their preimplantation scores.

Project description:IntroductionAlthough rare, cholesteatoma can develop as a late complication of cochlear implantation. The electrode array may then be exposed in the external auditory canal surrounded by cholesteatoma debris. Case Report. The cochlear implant of a child was inadvertently explanted by a clinician during a routine aural toilet procedure. The child had previously reported recurrent ear infections, pain, and unexplained implant function degradation. Reimplantation was carried out 2 days later with good postoperative hearing results. Part of the electrode array was observed to be embedded in cholesteatoma. Postreimplantation recovery was complicated by a breakdown of the blind-sac. Discussion. Clinical indicators that could alert the clinician to the possibility of this late complication include recurrent infections, presence of keratotic debris in the external auditory canal, unexplained implant function degradation, and nonauditory stimulation. Although this patient managed to achieve excellent postreimplantation hearing outcomes, a delay in reimplantation surgery following explantation could possibly compromise successful reinsertion of the electrode array. External ear canal overclosure without mastoid cavity obliteration has merit in facilitating CT scan surveillance, but it may increase the risk of the blind-sac breaking down. This case also illustrated how the electrode array could have facilitated propagation of the cholesteatoma from the middle ear to the mastoid.ConclusionIf aural toilet is required in the implanted ear of a cochlear implant recipient, any complaint of hearing change, pain, or discharge should alert the clinician of the possibility of cholesteatoma developing. It warrants prompt evaluation by an experienced otologist in order to prevent accidental explantation. Keywords. Cochlear implant, cochlear implant complications, chronic suppurative otitis media, cholesteatoma, reimplantation, blind-sac, external auditory canal overclosure, mastoid cavity obliteration.

Project description:Cochlear implantation is a surgical procedure, which is performed on severely hearing-impaired patients. Impedance field telemetry is commonly used to determine the integrity of the cochlear implant device during and after surgery. At the Department of Otolaryngology, Cheng Hsin General Hospital (Taipei, Taiwan), the cochlear implant devices are switched on within 24 hours of their implantation. In the present study, the impedance changes of Advanced Bionics™ cochlear implant devices were compared with previous studies and other devices. The aim was to confirm previous hypotheses and to explore other potential associated factors that could influence impedance following cochlear implantation. The current study included 12 patients who underwent cochlear implantation at Cheng Hsin General Hospital with Advanced Bionics cochlear implant devices. The cochlear devices were all switched on within 24 hours of their implantation. The impedance was measured and compared across all contact channels of the electrode, both intra-operatively and post-operatively. The intra-operative impedance was compared with the switch-on impedance (within 24 hours of the cochlear implantation); the impedance was notably increased for all contact channels at switch-on. Of the 16 channels examined, 4 channels had a significant increase in impedance between the intra-operative measurement and the switch-on measurement. To the best of our knowledge, the impedance of a cochlear implant device can be affected by the diameter of the electrode, the position of the electrode arrays in the scala tympani, sheath formation and fibrosis surrounding the electrode after implantation and electrical stimulation during or after surgery. When the results of the current study were compared with previous studies, it was found that the impedance changes were opposite to that of Cochlear™ implant devices. This may be explained by the position of the electrode arrays, sheath formation, the blow-out effect and differences in electrical stimulation.

Project description:ImportancePrevious research suggests that clinicians view realistic patient expectations as the most important nonaudiological factor in the decision to proceed with a cochlear implant (CI). However, clinicians have few data to determine whether patients' outcome expectations are realistic.ObjectiveTo address this unmet clinical need through the development and psychometric analysis of a new patient-reported outcome measure, the CI Quality of Life (CIQOL) Expectations.Design, setting, and participantsThis cross-sectional study was conducted at a tertiary CI center from February 26, 2020, to August 31, 2021. First, a team comprising 2 CI audiologists, a CI surgeon, a hearing scientist, and 2 psychometricians with experience in instrument development converted all items from the CIQOL-35 Profile instrument into statements reflecting expected outcomes. Then, cognitive interviews with 20 potential CI users assessed the clarity and comprehensiveness of the new instrument. Next, responses to the CIQOL-Expectations instrument for 131 potential adult CI candidates were psychometrically analyzed using confirmatory factor analysis and item response theory. Finally, degree to which patient expectations changed from before to after and their CI evaluation appointments was measured.InterventionThe CIQOL-Expectations instrument.ResultsOf 178 participants, 85 (47.8%) were female, and there was 1 (0.6%) Asian, 26 (14.6%) Black or African American, 1 (0.6%) Latinx, and 150 (84.3%) White individuals. No major content or grammar changes were identified during the cognitive interviews. Overall, all CIQOL domains demonstrated adequate to strong psychometric properties. Several domains did not meet all a priori established indicators of model fit or ability to separate CI users based on response patterns, but all met most indicators. Potential CI users demonstrated the highest mean (SD) expectation scores for the environment (70.2 [20.8]) and social (68.4 [18.0]) domains. In addition, the entertainment (20 [15.3%]) and environment (31 [24.4%]) domains had the highest percentage of patients with expectation scores of 100. Yet, normative CIQOL-35 Profile data from experienced CI users suggested few patients obtain this high degree of functional benefit after implant.Conclusions and relevanceThe results of this cross-sectional study suggest that the CIQOL-Expectations instrument may provide an opportunity to assess potential CI users' expected outcomes using modification of an established CIQOL instrument and a patient-centered framework. The included items and domains reflect real-world functional abilities valued by CI users and may provide opportunities for an evidence-based shared decision-making approach to the CI evaluation process. With this instrument, clinicians can compare individual patients' pre-CI outcome expectations with established normative data and provide appropriate counseling.

Dataset Information

A semi-supervised Support Vector Machine model for predicting the language outcomes following cochlear implantation based on pre-implant brain fMRI imaging.

Introduction

Methods

Results

Conclusion

Publications

A semi-supervised Support Vector Machine model for predicting the language outcomes following cochlear implantation based on pre-implant brain fMRI imaging.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets