Improving Accessibility and Usability of Clinical Data for the Swiss Personalized Health Network: Development and...

Area of Science:

• Health Informatics
• Medical Terminology
• Data Science

Background:

• Large-scale health research initiatives face challenges in data interoperability and standardization.
• The Swiss Personalized Health Network (SPHN) dataset integration presents complexities for researchers seeking specific clinical concepts.
• Semantic enrichment using SNOMED CT offers a solution for structured queries and improved data discoverability.

Purpose of the Study:

• To evaluate the impact of a semantic layer on exploring the SPHN dataset using SNOMED CT.
• To develop and validate the Smart SNOMED Search for SPHN (S4) tool for semantic data exploration.
• To facilitate structured semantic searches via SNOMED CT's Expression Constraint Language.

Main Methods:

• Systematic mapping of SPHN dataset concepts and attributes to SNOMED CT codes and value sets.
• Development of the S4 tool for semantic enrichment and Expression Constraint Language-based queries.
• Validation of the S4 tool using a clinical data warehouse dataset, assessing precision, recall, and F1-scores.

Main Results:

• The S4 tool achieved high accuracy: 95.3% precision, 97.5% recall, and 96.4% F1-score.
• Semantic enrichment identified gaps in the SPHN dataset, such as lack of representation and postcoordination.
• Enhanced semantic connections improved data discoverability and alignment with SNOMED CT.

Conclusions:

• Semantic representation significantly enhances data explorability and accessibility within large health frameworks like SPHN.
• The S4 tool effectively validates the benefits of semantic enrichment for research.
• Future work should focus on refining search precision and user accessibility to further support personalized health research.