Ontologies in Big Health Data Analytics: Application to Routine Clinical Data.

Area of Science:

• Health Informatics
• Big Data Analytics
• Clinical Data Management

Background:

• Traditionally, code lists were manually created by domain experts and mapped between systems.
• Existing methods for extracting and analyzing clinical data present challenges in transparency and semantic resolution.
• Healthcare data is often fragmented across different coding systems and general practice systems.

Purpose of the Study:

• To operationalize ontologies as a tool for big-data analytics in healthcare research.
• To improve the representation and extraction of complex clinical data.
• To demonstrate a transparent method for case identification and outcome measurement using ontologies.

Main Methods:

• Utilizing ontologies to resolve semantic complexities within diverse healthcare datasets.
• Developing and applying an ontology-driven process for data extraction and analysis.
• Demonstrating the method's efficacy through a pregnancy case identification use case.

Main Results:

• Ontologies provide a robust framework for representing clinical concepts and data.
• The operationalized process enhances transparency in case identification and outcome measurement.
• The pregnancy case identification method effectively handles complexities such as varying lengths and multiple care providers.

Conclusions:

• Ontologies are valuable tools for advancing big-data analytics in healthcare.
• The described method offers a transparent and efficient approach to analyzing complex clinical data.
• Ontology-driven approaches facilitate better data extracts and semantic resolution for research.