Employing complex polyhierarchical ontologies and promoting interoperability of i2b2 data systems

Area of Science:

• Biomedical Informatics
• Data Management
• Ontology Engineering

Background:

• i2b2 is widely used for research data warehouses, utilizing reference ontologies and PATH-based queries.
• Large, polyhierarchical ontologies like SNOMED CT present dissemination and deployment challenges within i2b2.
• Current i2b2 search methods based on node traversal (PATHs) can be complex and inefficient.

Purpose of the Study:

• To evaluate an alternative approach using transitive closure tables (TC) for i2b2 data management.
• To compare the search speed, accuracy, and interoperability of TC-based versus PATH-based queries.
• To determine if TC tables can improve the efficiency of using SNOMED CT within i2b2.

Main Methods:

• Implemented and evaluated both TC-based and PATH-based query approaches within the i2b2 framework.
• Measured search speed and accuracy for queries involving concepts within large ontologies.
• Analyzed Oracle query plan resource estimates to quantify performance differences.

Main Results:

• Both TC-based and PATH-based queries demonstrated comparable accuracy.
• TC-based queries were substantially faster than PATH-based queries for concepts present in numerous paths.
• Query resource estimates showed one to three orders of magnitude difference, favoring TC tables.

Conclusions:

• Transitive closure tables offer a more efficient and interoperable method for i2b2 data searching compared to PATH-based approaches.
• Simplifying i2b2 metadata build and dissemination tools can effectively leverage SNOMED CT with TC tables for increased efficiency.
• Implementing TC tables in metadata can significantly enhance network query interoperability for research data warehouses.