One shared data repository has supported five research studies at Charité instead of a separate infrastructure for every project. A new paper accepted on October 6 describes an environment covering about 16,000 participants and 7.9 million data points.
The problem is familiar to research software teams. Different projects repeatedly translate between data formats, build their own capture systems and accumulate definitions that no longer match. That makes later reuse harder even when information is technically available.
The proposed environment stores data using openEHR, a standard that separates reusable information models from the applications that handle them. Governed templates provide a common meaning for stored values.
The researchers connect that repository with FHIR, a standard for exchanging health information, and OMOP, a common structure for analytics. The architecture combines existing records, study data, reported outcomes and sensor streams.
Of 125 information models, the authors say 57 are shared by at least two studies. That is evidence of reuse within the reported deployment, rather than proof that any institution can adopt the system without work.
The paper also preserves an important limit. Direct study capture removes one separate research-entry system, but duplicate documentation remains where the hospital's main information system does not use openEHR.
Tests at three institutions demonstrated cross-site cohort-count queries. Those federation nodes are not yet in routine use, according to the authors.
The tooling is public. Sustainability beyond the five studies and scaling beyond one institution remain to be demonstrated.
