Twenty-five years of clinical data collection: from a single site relational database towards multi-site interoperability

Author: Eugen Kludt & Andreas Büchner, Department of Otolaryngology, Medical University of Hannover, Hanover, Germany

The Hannover Medical School (Medizinische Hochschule Hannover, MHH), the world’s largest center for cochlear implants (CI) and implantable hearing aids, has implanted more than 10 000 CI devices since the first implantation in 1984. Growing numbers of patients and collection of their clinical performance data during the life-long postoperative care made efficient clinical data management necessary. This led to an early application of a relational database system to store demographics, anamnesis and speech intelligibility data, becoming operational in 1996. Widening of CI candidacy to patients with low frequency residual hearing and ongoing advances in digitization, increased the number of relevant data sources (e.g. digital audiometers, imaging, fitting data of hearing devices or smartphone based remote care). Information on one patient, collected at different places within the clinic had to be joined together in order to access all individual parameters that might be relevant for the success of the CI implantation. Over the years, this has led to a comprehensive data pool for patients with implantable hearing devices, which serves as a basis for answering various research questions and big data analyses. A logical next step would be the pooling of data across sites, i.e. joining of data from different clinics to tackle unresolved research questions needing larger amounts of data. To achieve this task, the data formats between different sites need to be unified – at least at the level of an across-site data warehouse. Open standards for audiological measurements, fitting parameters and related data will most likely facilitate the unification of data storage as well as data collection across sites – thereby reducing the hurdles to pool data from different centers for future big data machine learning projects.