Neuro-AI guided objective hearing assessment and hearing loss compensation – Neurogram

Wider research context: Hearing loss is pervasive and significantly impacts daily life. Untreated hearing loss in middle age is recognized as the “largest modifiable risk factor“ for dementia. Thus, accurately quantifying hearing damage is crucial, and the Pure Tone Audiogram (PTA) remains the gold standard. However, PTA's scope is restricted to detecting faint pure artificial tones, offering limited insights into the processing of supra-threshold sounds. Also, existing methods for hearing assessment lack objective approaches to optimally compensate for hearing loss, resulting in frequently unsatisfactory experiences with hearing aids.

Objectives: To address the aforementioned issues by implementing and investigating a novel approach. We have recently devised and tested a method labeled “Neurogram“in a comprehensive pilot study. Building on this foundation, our objective is to evaluate whether the Neurogram can narrow the divide between objective and subjective hearing assessments. Given that the initial Neurogram was developed using MEG, we aim to assess the feasibility of employing EEG, a more widely available method. Crucially, the project will incorporate an AI-based approach, enabling us to predict the impacts of sound transformations on the Neurogram.

Approach: The project builds upon the pilot study. At the heart of the approach is the reconstruction of acoustic features in defined sound frequency bands, from electrophysiological data recorded while participants listen to engaging audiobooks. Next to a more conventional regression-based approach, we will develop a novel approach using Deep Neural Networks that enables prediction of Neurogram changes by changes of sounds using a hearing aid simulator. We will obtain objective (PTA, ABR, TEN-, Speech-In-Noise Test) and subjective (SSQ) parameters of hearing. Most participants will be recruited from an ongoing online study (“Wie ho¨rt Salzburg?”).

Innovation: For over a century, the primary method for assessing hearing, namely PTA, has remained largely unaltered, despite acknowledged shortcomings. While alternative approaches exist that address some of these issues, none have successfully challenged the entrenched position of PTA as the gold standard. The Neurogram presents the most promising potential for comprehensively overcoming both the scientific and practical limitations of PTA. This ongoing project holds the potential for a transformative impact on the future assessment of hearing status and serves as an objective foundation for improved sound transformations increasing listening experiences of hearing aid users.

Primary researchers involved: The Austrian scientists that have developed and piloted the Neurogram at the University of Salzburg (PI: Nathan Weisz; Co-PI: Fabian Schmidt) will advance the project together with a French AI specialist (PI: Timothée Masquelier) at the CNRS CERCO. This core team will collaborate with Torsten Dau (DTU) and Nicolas Lesica (UCL Ear Institute).