Classification of Acoustic Hearing Preservation After Cochlear Implantation Using Electrocochleography.
Leanne Sijgers, Torquil Sorensen, Andrew Soulby and 9 others
PMID 38105510WHAT IT FOUND
Intraoperative electrocochleography failed to predict which cochlear implant patients would lose residual hearing.
Neither the drop in the difference response nor machine learning classifiers trained on these signals could distinguish between patients who preserved hearing and those who lost it.
Key findings
01Neither overall nor relative decreases in the DIF response amplitude could distinguish between acoustic hearing loss and hearing preservation in this cohort.
02Support vector machine and random forest classifiers trained on features extracted from intraoperative electrocochleography recordings could not distinguish cases with hearing loss and hearing preservation.
03Exploratory analysis of the feature set suggested that the features were not discriminating well between the classes with hearing preservation and hearing loss.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The sample size was small (40 subjects), which limits the statistical power and generalizability of the machine learning models. ECochG recordings were analyzed in steps (pseudo-stepwise or actual stepwise) rather than continuous real-time analysis, which may have missed critical transient changes in the signal. Two different types of electrode arrays (SlimJ and Mid-Scala) were used, which may have different trauma mechanisms, introducing heterogeneity in the data. Some hearing loss may have occurred in the postoperative period due to factors like fibrosis or electrode migration, which intraoperative ECochG cannot detect. The features selected for classification were based on prior research and may have been suboptimal for capturing the complex relationships between ECochG signals and hearing outcomes.
Declared interests
The study was partially funded by Advanced Bionics, the manufacturer of the cochlear implant devices used, and InnoSuisse. The authors declared no other potential conflicts of interest.
The easy way to misread this
Do not conclude that intraoperative electrocochleography is useless for monitoring cochlear trauma. This study tested a specific set of features from stepwise recordings in a small cohort. The failure to predict hearing loss here does not negate previous studies showing correlations between ECochG changes and hearing outcomes, but it does highlight that current real-time feedback methods are not yet reliable enough to guarantee hearing preservation.