Data-Driven Approach for Auditory Profiling and Characterization of Individual Hearing Loss

Area of Science:

• Audiology
• Hearing Science
• Signal Processing

Background:

• Pure-tone audiometry is standard for hearing loss characterization and hearing aid fitting.
• Audiograms often fail to capture perceptual clarity deficits, impacting real-world listening.
• A simplified approach is needed to assess individual hearing compensation requirements.

Purpose of the Study:

• To hypothesize and validate a two-dimensional model (Type I and Type II distortions) for characterizing auditory profiles.
• To develop a classification system for identifying distinct listener profiles based on hearing distortion.
• To enhance strategies for hearing deficit compensation.

Main Methods:

• Reanalysis of existing individual performance data from two studies.
• Application of unsupervised learning to identify extreme patterns and auditory profiles.
• Development of a decision tree for listener classification into identified profiles.

Main Results:

• Evidence found for four distinct auditory profiles within the analyzed data.
• Key predictors for profile identification included binaural processing, auditory nonlinearity, and speech-in-noise perception.
• The proposed two-dimensional model effectively categorizes listeners based on distortion types.

Conclusions:

• A novel framework characterizing hearing loss via two distortion dimensions (Type I and Type II) is supported.
• The identified profiles offer a simplified yet comprehensive approach to auditory profiling.
• This method can refine hearing aid fitting and compensation strategies by considering perceptual clarity alongside audibility.