Dotting the “i” of Interoperability in FAIR Cancer-Registry Data Sets

Abstract

To conform to FAIR principles, data should be findable, accessible, interoperable, and reusable. Whereas tools exist for making data findable and accessible, interoperability is not straightforward and can limit data reusability. Most interoperability-based solutions address semantic description and metadata linkage, but these alone are not sufficient for the requirements of inter-comparison of population-based cancer data, where strict adherence to data-rules is of paramount importance. Ontologies, and more importantly their formalism in description logics, can play a key role in the automation of data-harmonization processes predominantly via the formalization of the data validation rules within the data-domain model. This in turn leads to a potential quality metric allowing users or agents to determine the limitations in the interpretation and comparability of the data. An approach is described for cancer-registry data with practical examples of how the validation rules can be modeled with description logic. Conformance of data to the rules can be quantified to provide metrics for several quality dimensions. Integrating these with metrics derived for other quality dimensions using tools such as data-shape languages and data-completion tests builds up a data-quality context to serve as an additional component in the FAIR digital object to support interoperability in the wider sense.

An ontology design for validating childhood cancer registry data

“…The ontology had to be redesigned to allow a more scalable and comprehensive approach to the rules and to build on a number of shared core ontologies. Two validation modules dealing with cancer stage (7) and multiple primary tumours (8) have been developed according to this principle. Both these ontologies were developed as stand-alone applications since they are computationally quite demanding tasks and generally apply only to a subset of cancer registry case records, but they draw on the same shared core ontologies.…”

Ontologies as a Tool for Formalizing Data Validation Rules

“…Identification of the high-level data concepts required to answer those questions In this breakdown, the process of data validation falls manly under steps 5) and 6) although it should be stressed that validation forms only part of the quality-control procedures of step 6). Other fundamental quality metrics consist of the following dimensions: completeness, consistency, accuracy, timeliness, uniqueness, and auditability [27]. Moreover, different entities in the data process may be responsible for 6 and 7.…”

Ontology-Based AI Design Patterns and Constraints in Cancer Registry Data Validation

“…Application of a consistent and standard approach has been one of the hurdles in the past since it is critical that no biases are introduced from an inconsistent application of the validation rules. A further advantage of the ontological approach is that it allows some quantifiable and reproducible means of measuring several quality metrics of data sets, affording users greater assurance in the comparability/integration of data sets from different sources [39].…”