Knowledge Driven Phenotyping.

Area of Science:

• Health Informatics
• Clinical Research Informatics
• Data Science in Healthcare

Background:

• Extracting patient phenotypes from electronic health records (EHRs) is crucial for clinical research.
• Current methods involve translating phenotype definitions into executable queries, which is complex and data-dependent.
• These translations are often time-consuming, error-prone, and lack reproducibility across diverse healthcare settings.

Purpose of the Study:

• To propose a novel knowledge-driven framework for phenotype extraction.
• To decouple phenotype semantics from specific data sources, enabling broader applicability.
• To automate the population and execution of phenotype computations across heterogeneous data.

Main Methods:

• Development of a knowledge-driven framework separating phenotype definition from data.
• Implementation of automated processes for populating and conducting phenotype computations.
• Deployment and evaluation of the framework on five distinct Scottish health datasets.

Main Results:

• The framework successfully decouples phenotype semantics from underlying data structures.
• Automated computation of phenotypes was achieved across heterogeneous data spaces.
• Preliminary results demonstrate the feasibility and potential of the framework in real-world health data.

Conclusions:

• The proposed framework offers a more efficient, reproducible, and shareable approach to patient phenotype extraction.
• This methodology can significantly reduce the technical burden on clinical researchers.
• Further validation and expansion of the framework to other health datasets are warranted.