Toward cross-platform electronic health record-driven phenotyping using Clinical Quality Language

Area of Science:

• Biomedical Informatics
• Health Data Science
• Clinical Decision Support

Background:

• Electronic health record (EHR)-driven phenotyping is crucial for biomedical knowledge discovery.
• Current phenotyping methods are manual, slow, error-prone, and platform-specific, hindering scalability and portability.
• The nascent Clinical Quality Language (CQL) offers a potential solution to these limitations.

Purpose of the Study:

• To investigate the utility of Clinical Quality Language (CQL) for high-throughput, cross-platform EHR phenotyping.
• To assess CQL's expressiveness in representing complex clinical phenotype definitions.
• To evaluate the performance and portability of a CQL-based phenotyping approach.

Main Methods:

• Translated a validated heart failure (HF) phenotype definition into CQL.
• Developed a CQL execution engine integrated with the Observational Health Data Sciences and Informatics (OHDSI) platform.
• Executed the CQL phenotype definition at two academic medical centers and verified results for precision and recall.
• Performed cross-platform execution against OHDSI and Fast Healthcare Interoperability Resources (FHIR) data platforms.

Main Results:

• CQL effectively represents complex HF phenotype definitions, including temporal relationships.
• The custom CQL execution engine was implemented with relative ease.
• Results verification demonstrated 100% precision and 100% recall for the HF phenotype.
• Cross-platform execution using CQL yielded identical patient cohorts across OHDSI and FHIR platforms.

Conclusions:

• CQL is capable of representing complex clinical phenotype definitions.
• The developed CQL execution engine enables cross-platform phenotyping against major data platforms.
• CQL has the potential to significantly improve portability and scalability in EHR-driven phenotyping for learning health systems.