RA 3: Data-driven personalization of hearing support

To address the challenges of data heterogeneity and incompatibility on an international scale, we have made data standardization a central aspect of our new mission to improve data exchange, communication, and interoperability.

With our Hearing4all database, containing longitudinal data from more than 10,000 patients over a period of up to 30 years, we are now spearheading the first international efforts to harmonize audiological data. Utilizing openEHR standards, we aim to achieve seamless data sharing among stakeholders and establish an international Open Data Space for Audiology.

With the fundamental expertise of our new PIs, our ambition extends to crafting common information models as computable, formal representations of domain data encompassing data structures and metadata – in standardized formats. This ambitious project aims not only to facilitate sharing and analysis of data beyond the Cluster, but also to foster an environment of unprecedented international collaborative research and innovation. As a fundamental part of this initiative, we contribute our very extensive, existing datasets to this public domain while preserving data privacy and security, reflecting our belief in the transformative power of shared knowledge and data.

Drawing on recent successes in AI, where systems have significantly benefited from learning representations from large datasets, we leverage advanced data-driven approaches to representation learning to enhance the patients’ generalized auditory profile. The data-driven profiles closely align with hearing loss causes, and will be used for the development of routines that estimate disease probabilities for AI-informed diagnostics. Furthermore, we develop a Decision Support System (DSS), that uses the AI-informed estimation of hearing cause probabilities, and integrates treatment suggestions derived from data of the Open Data Space and learned data representations. The DSS is primarily intended for clinical settings, working with comprehensive patient data, but will also be accessible via our Virtual Hearing Clinic (VHC), providing data outside of clinical consultations.

Together, the VHC and DSS will yield unprecedented insights into patient conditions, essential for creating personalized and effective hearing solutions.