Simulating speech processing with cochlear implants: How does channel interaction affect learning in neural networks?

Predicting Acoustic Hearing Preservation Following Cochlear Implant Surgery Using Machine Learning

OBJECTIVES

The aim of the study was to train and test supervised machine-learning classifiers to predict acoustic hearing preservation after CI using preoperative clinical data.

STUDY DESIGN

Retrospective predictive modeling study of prospectively collected single-institution CI dataset.

METHODS

One hundred and seventy-five patients from a REDCap database including 761 patients >18 years who underwent CI and had audiometric testing preoperatively and one month after surgery were included. The primary outcome variable was the lowest quartile change in acoustic hearing at one month after CI using various formulae (standard pure tone average, SPTA; low-frequency PTA, LFPTA). Analysis involved applying multivariate logistic regression to detect statistical associations and training and testing supervised learning classifiers. Classifier performance was assessed with numerous metrics including area under the receiver operating characteristic curve (AUC) and Matthews correlation coefficient (MCC).

RESULTS

Lowest quartile change (indicating hearing preservation) in SPTA was positively associated with a history of meningitis, preoperative LFPTA, and preoperative SPTA. Lowest quartile change in SPTA was negatively associated with sudden hearing loss, noise exposure, aural fullness, and abnormal anatomy. Lowest quartile change in LFPTA was positively associated with preoperative LFPTA. Lowest quartile change in LFPTA was negatively associated with tobacco use. Random forest demonstrated the highest mean classification performance on the validation dataset when predicting each of the outcome variables.

CONCLUSIONS

Machine learning demonstrated utility for predicting preservation of residual acoustic hearing in patients undergoing CI surgery, and the detected associations facilitated the interpretation of our machine-learning models. The models and statistical associations together may be used to facilitate improvements in shared clinical decision-making and patient outcomes.

LEVEL OF EVIDENCE

3 Laryngoscope, 134:926-936, 2024.

Is the spread of excitation width correlated to the speech recognition in cochlear implant users?

PURPOSE

To assess whether there is an interference of the spread of excitation (SOE) on speech recognition.

METHODS

Retrospective cross-sectional study, approved by the institution's ethics committee (CAAE03409212.8.0000.0068). Adult patients with intraoperative neural response telemetry (NRT) performed on electrodes 6, 11 and 16 implanted with Cochlear Ltd (Sydney, Australia) devices were selected. Patients with partial array insertion, pre-lingual hearing loss, deafness etiology due to and CI experience less than 12 months were excluded. SOE was recorded at 10 current units above the NRT threshold (tNRT) and its width in millimeters was collected at point 0.75 of the function. Speech recognition test was 25-recorded monosyllables list, presented at 65 dBHL at 0° azimuth in a sound treated booth. The analysis was divided into groups by electrode array type, regarding the tNRT, SOE width, SOE's peak amplitude and electrode peak.

RESULTS

A 126 SOE measurements of the 3 tested electrodes were obtained from 43 patients. Patients with straight array had significantly wider SOE, greater peak amplitude at electrode 6 and higher tNRTs. In the perimodiolar array, there was a negative correlation between SOE and monosyllables recognition at electrodes 6 and 11, and in the combined average of the three electrodes, with a significant difference in electrode 11. Sixty-six percent of the SOE measurements had their peak shifted to adjacent electrodes.

CONCLUSION

It was observed, in perimodiolar array, the greater the dispersion of electrical current, the worse the speech recognition, especially in the medial electrode.