Transforming cancer registry data into an ontology-driven common data model: Evaluation of the implementation and integrability of the Observational Medical Outcomes Partnership Model in Clinical Cancer Registry

Introduction: The identification and incorporation of biomarkers in the treatment of people suffering from cancer is moving from guideline-based treatment towards a personalized approach in cancer care. This individual approach in cancer medicine, requires very granular inclusion criteria to conduct a study, minimizes the number of potential study participants resulting in limited power of the study. Therefore, cross-institutional data sharing to conduct collaborative studies in cancer research is becoming increasingly important. Semantic heterogeneity regarding data representation within different research institutions complicates data exchange. A possible solution could be the embedding of individualized cancer data into a common data model, which represents knowledge with a unified semantic. Methods: Cancer registry data from the UCCH were adapted to the standardization of the OMOP CDM v.6.0. In addition to the implementation, the CDM was tested with regard to two use cases. The first use case examines to which extent the ADT/GEKID data set can be integrated into a CDM for cross-institutional data exchange. The second use case examines if key performance indicators of the urinary bladder module within the scope of organ centre certification, with the goal to ensure quality of care in hospital, can be represented using CDM. Results: A total of 1936055 data entries from the source system could be implemented as standardized concepts in OMOP CDM v.6.0. 80626 data entries were implemented as custom vocabularies. Through the ontologies, which are already integrated in the OMOP CDM, 78.5% of the ADT/GEKID base data set can be mapped. When determining the key performance indicators of bladder module in the context of organ centre certification, the CDM seems to slightly underestimate when including the arithmetical mean (x̅ = -0.5). Conclusions: The OMOP CDM v.6.0 is a suitable tool for data harmonization in operational databases with heterogeneous semantic. For the transmission of cancer data within the scope of ADT/GEKID, the model is suitable for 78.5% of the data. The transfer of the ADT/GEKID into the vocabulary of the OMOP CDM could contribute a big step in European data integration. For the determination of organ-specific key performance indicators in the context of a certification, the model is suitable. However, the vocabulary of the CDM have to be extended for this use case