An eligibility criteria query language for heterogeneous data warehouses

Area of Science:

• Health Informatics
• Clinical Research Informatics
• Data Interoperability

Background:

• Electronic clinical data offers research potential but faces challenges due to data heterogeneity.
• Existing query languages struggle with diverse data representations, hindering consistent interrogation.
• A need exists for a higher-level, domain-specific language for data-representation-agnostic queries.

Purpose of the Study:

• To define a clinician-readable computational language for querying patient eligibility criteria (ECs) from clinical trials.
• To enable automatic evaluation of ECs on heterogeneous data sources, abstracting data complexity.
• To overcome limitations of standard query languages in diverse clinical data environments.

Main Methods:

• Developed a clinician-readable query language based on existing EC standards and examples.
• Implemented a model-based approach to transform ECs into queries for heterogeneous data warehouses.
• Evaluated the query language on two distinct data sources with differing structures and content.

Main Results:

• The query language abstracts expressivity, allowing EC construction without prior data source knowledge.
• Successfully evaluated on semantically and structurally diverse data warehouses.
• The language is utilized in the EHR4CR project and perceived as useful, understandable, and unambiguous by users.

Conclusions:

• An EC-specific language isolates researchers from data complexities, enabling effective querying of heterogeneous clinical data.
• The approach demonstrates the potential of domain query languages for overcoming semantic interoperability challenges.
• The developed language is usable, interoperable, and maintains semantic consistency across clinical domains.