Index-based measure for hearing capabilities and hearing device indication criteria: Enabling model-based comparison across speech tests in European clinical databases

• Applicant: Dr. Mareike Buhl
• Subject Area: Acoustics
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 496819293

Hearing devices aim to restore hearing loss which cannot be treated, and to provide the optimal device benefit for each patient such that speech communication in everyday listening situations is enabled. The choice of hearing devices for individual patients is formalized in clinical guidelines, where indication criteria for hearing aids and cochlear implants are defined. These criteria are mainly based on audibility and speech recognition, however, the influence of additional factors on benefit of hearing devices is still subject of research.Analyses and predictions based on large amounts of data as stored in clinical databases could support the investigation of important factors for optimizing indication criteria, as well as contribute to personalized medicine by providing recommendations for different hearing devices to support the decision-making of clinical experts. However, local clinical databases differ in terms of included measurements and information, especially regarding speech tests. Differences across countries comprise speech tests performed in the respective language, but also different complexity and conditions. An objective measure for comparability of speech tests is missing, but could be achieved with speech recognition models.Therefore, the main aim of the current project is to enable comparability of clinical-audiological databases across countries. For this purpose, a new approach for comparability of speech tests will be developed, combined with the previously developed approach of Common Audiological Functional Parameters (CAFPAs), and applied to French and German databases to obtain a proof of concept for comparability of databases based on two countries.After organizing access to the databases, the first sub-goal is the development of a hearing loss index which describes the effective hearing impairment of a patient in an interval between 0 and 1. The index will be independent of the employed speech test as it will be derived from the audibility (A) and distortion (D) component of the Plomp (1978) concept, which can in turn be estimated from a combination of available measurements and speech recognition model predictions of the respective speech test.The second sub-goal is to combine the hearing loss index with the previously developed CAFPAs, which are abstract parameters independent of the choice of measurements and should therefore provide comparability across databases. The CAFPAs have previously been evaluated for one speech test, and are now intended to be estimated from different speech tests by including a translation via the hearing loss index.Third, comparability of clinical databases will be assessed using the developed methods, by investigating and comparing the hearing device indication criteria and benefit in the hearing loss index domain, and performing a classification based on these criteria.Finally, future collaborations to test the feasibility of the developed approaches will be organized.

• DFG Programme: WBP Fellowship
• International Connection: France
• Host: Professor Dr. Paul Avan