Clinical Auditory Profiling: Divide and Conquer

By Raul Sanchez-Lopez and Torsten Dau

This article is a part of the July/August 2022, Volume 34, Number 4, Audiology Today issue. 

For years, several approaches attempted to disentangle the physiological and perceptual consequences of different hearing loss etiologies (Johannesen et al, 2016; Lecluyse et al, 2013; Lopez-Poveda and Johannesen, 2012; Strelcyk and Dau, 2009). Despite these efforts, hearing deficit characterization and rehabilitation are currently still dominated by the hearing thresholds obtained from pure-tone audiometry. Adoption of complementary measures of auditory function, such as speech-in-noise tests, faces barriers of clinical implementation, and the use of such additional information is not clearly linked to specific hearing rehabilitation interventions (Anderson et al, 2018). In other disciplines, such as genetics or business analytics, profiling has recently gained attention as a tool for typifying groups of observations that follow similar patterns. In profiling, extensive information is collected, enabling the discovery of meaningful associations across, for example, genetic profiles and medical conditions, through computational analysis (Li et al, 2004). Clinical biomarkers that enable the stratification of patients in relevant subpopulations can, then, be discovered and implemented in clinical practice (Trusheim et al, 2007). In our recent work, we explored the possibility of using data-driven analysis to identify archetypal patterns among a population of patients with hearing loss. The goal was to (1) identify subpopulations with distinct differences in hearing abilities across the subpopulations; (2) make this stratification method interpretable and meaningful, such that it can easily be applied to drive decisions in hearing rehabilitation; and (3) propose a “precision audiology” paradigm, where the listeners are classified into auditory profiles and the hearing aid fitting process is optimized based on their specific hearing deficits beyond the audiogram.