The Collaboration between Hearing Aid Users and Artificial Intelligence to Optimize Sound

Artificial Intelligence in Otology and Neurotology

Clinical applications of artificial intelligence (AI) have grown exponentially with increasing computational power and Big Data. Data rich fields such as Otology and Neurotology are still in the infancy of harnessing the power of AI but are increasingly involved in training and developing ways to incorporate AI into patient care. Current studies involving AI are focused on accessible datasets; health care wearables, tabular data from electronic medical records, electrophysiologic measurements, imaging, and "omics" provide huge amounts of data to utilize. Health care wearables, such as hearing aids and cochlear implants, are a ripe environment for AI implementation.

A Review of Machine Learning Approaches for the Personalization of Amplification in Hearing Aids

This paper provides a review of various machine learning approaches that have appeared in the literature aimed at individualizing or personalizing the amplification settings of hearing aids. After stating the limitations associated with the current one-size-fits-all settings of hearing aid prescriptions, a spectrum of studies in engineering and hearing science are discussed. These studies involve making adjustments to prescriptive values in order to enable preferred and individualized settings for a hearing aid user in an audio environment of interest to that user. This review gathers, in one place, a comprehensive collection of works that have been conducted thus far with respect to achieving the personalization or individualization of the amplification function of hearing aids. Furthermore, it underscores the impact that machine learning can have on enabling an improved and personalized hearing experience for hearing aid users. This paper concludes by stating the challenges and future research directions in this area.

Mechanical effect of reconstructed shapes of autologous ossicles on middle ear acoustic transmission

Conductive hearing loss is caused by a variety of defects, such as chronic otitis media, osteosclerosis, and malformation of the ossicles. In such cases, the defective bones of the middle ear are often surgically reconstructed using artificial ossicles to increase the hearing ability. However, in some cases, the surgical procedure does not result in increased hearing, especially in a difficult case, for example, when only the footplate of the stapes remains and all of the other bones are destroyed. Herein, the appropriate shapes of the reconstructed autologous ossicles, which are suitable for various types of middle-ear defects, can be determined by adopting an updating calculation based on a method that combines numerical prediction of the vibroacoustic transmission and optimization. In this study, the vibroacoustic transmission characteristics were calculated for bone models of the human middle ear by using the finite element method (FEM), after which Bayesian optimization (BO) was applied. The effect of the shape of artificial autologous ossicles on the acoustic transmission characteristics of the middle ear was investigated with the combined FEM and BO method. The results suggested that the volume of the artificial autologous ossicles especially has a great influence on the numerically obtained hearing levels.

Preference for Combinations of Hearing Aid Signal Processing

PURPOSE

The purpose of this study was to determine how multiple types of signal processing activated together influence listeners' preferences.

METHOD

Participants were adults with mild to moderately severe sensorineural hearing loss. Stimuli were spatialized low-context sentences mixed with six-talker babble at 3 and 8 dB signal-to-noise ratios (SNRs). Stimuli were processed with three common hearing aid processing algorithms: wide dynamic range compression (WDRC), frequency compression (FC), and digital noise reduction (DNR). A full-factorial design with two levels for each algorithm (WDRC & DNR: mild versus strong; FC: ON versus OFF; clinically relevant ranges) was evaluated. Preference was measured using a paired-comparison task within a choice-based conjoint analysis framework. Remote data collection methods were used. A signal fidelity metric quantified the acoustic effects across conditions.

RESULTS

At 3 dB SNR, participants preferred a combination of Slow WDRC and Mild DNR, although the mean preference was small (odds ratio close to 1). At both SNRs when Strong DNR was used, Fast WDRC was preferred over Slow WDRC. This may be related to signal fidelity, which was lower for the combination of Fast WDRC and Mild DNR and higher for the combination of Slow WDRC and either Mild DNR or Strong DNR. There was no effect of FC on preference or signal fidelity.

CONCLUSIONS

WDRC and DNR together influenced both listeners' preferences and signal fidelity in the investigated listening conditions. On average, the small effect sizes suggest that minor fine-tuning adjustments to hearing aid algorithms may not result in a substantial change in clinical outcomes.

Translational Applications of Machine Learning in Auditory Electrophysiology

Machine learning (ML) is transforming nearly every aspect of modern life including medicine and its subfields, such as hearing science. This article presents a brief conceptual overview of selected ML approaches and describes how these techniques are being applied to outstanding problems in hearing science, with a particular focus on auditory evoked potentials (AEPs). Two vignettes are presented in which ML is used to analyze subcortical AEP data. The first vignette demonstrates how ML can be used to determine if auditory learning has influenced auditory neurophysiologic function. The second vignette demonstrates how ML analysis of AEPs may be useful in determining whether hearing devices are optimized for discriminating speech sounds.

Ecological Momentary Assessment to Obtain Signal Processing Technology Preference in Cochlear Implant Users

Background: To assess the performance of cochlear implant users, speech comprehension benefits are generally measured in controlled sound room environments of the laboratory. For field-based assessment of preference, questionnaires are generally used. Since questionnaires are typically administered at the end of an experimental period, they can be inaccurate due to retrospective recall. An alternative known as ecological momentary assessment (EMA) has begun to be used for clinical research. The objective of this study was to determine the feasibility of using EMA to obtain in-the-moment responses from cochlear implant users describing their technology preference in specific acoustic listening situations. Methods: Over a two-week period, eleven adult cochlear implant users compared two listening programs containing different sound processing technologies during everyday take-home use. Their task was to compare and vote for their preferred program. Results: A total of 205 votes were collected from acoustic environments that were classified into six listening scenes. The analysis yielded different patterns of voting among the subjects. Two subjects had a consistent preference for one sound processing technology across all acoustic scenes, three subjects changed their preference based on the acoustic scene, and six subjects had no conclusive preference for either technology. Conclusion: Results show that EMA is suitable for quantifying real-world self-reported preference, showing inter-subject variability in different listening environments. However, there is uncertainty that patients will not provide sufficient spontaneous feedback. One improvement for future research is a participant forced prompt to improve response rates.