Authors: Lena Schell-Majoor¹; Mareike Buhl¹ ²; Eugen Kludt³; Daniel Berg¹; Birger Kollmeier¹; Antje Wulff¹
¹Carl von Ossietzky Universität Oldenburg; ²Institut de l’Audition and IHU reConnect; ³Medizinische Hochschule Hannover
Background: Analyses of combined, diverse audiological data from multiple locations are becoming increasingly important, for example in the context of AI-based decision support systems, or multi-center studies. To facilitate cross-institutional exchange, the efficient use of data and the combination of different data sources, standardization of audiological data is essential. An EFAS working group is working on the standardization of audiological data to advance data-driven audiology. The critical issue is data modelling, i.e., to determine which parameters need to be documented in which way in order to fully “understand” any new data set in a comprehensive, expert-knowledge-driven way that will allow to convert the provided data set into any standardized data format.
Method: openEHR is an open interoperability standard that allows medical concepts to be described and stored in an unambiguous syntactic and semantic manner. Audiological tests are modelled using so-called archetypes, which can later be combined to templates to suit various use cases. Archetypes define which data elements are relevant to the respective clinical concept and describe the available data and metadata as completely as possible. To evaluate and optimize content and structure of the archetypes, a review was conducted with the audiological community.
Results: The specifities and output of the specific audiological modelling process are presented and discussed. We show the current audiological openEHR archetypes for loudness scaling, speech tests, and audiogram, which are published, under revision, and under development, respectively.
Conclusion: In summary, openEHR archetypes are proposed for three typical audiological tests and can be further refined through additional community review input. It is important to incorporate perspectives from clinical practice, research, and data structures even during the modelling phase, to achieve widely accepted data standards.
