Semantically enabling clinical decision support recommendations

Area of Science:

• Medical Informatics
• Biomedical Ontologies
• Knowledge Representation

Background:

• Clinical decision support systems (CDSS) use evidence-based recommendations from guidelines.
• Current CDSS often lack transparency, relevance, and applicability for clinicians.
• Previous technical approaches did not focus on formalizing guideline changes or evidence provenance.

Purpose of the Study:

• To enhance CDSS capabilities using semantic techniques.
• To model diseases, guideline provenance, and study cohorts.
• To improve adaptability to guideline changes and support personalized explanations.

Main Methods:

• Developed ontologies and semantic web tools for guideline modeling, provenance, and study cohort modeling.
• Utilized a custom-built knowledge graph framework unified by standard biomedical ontologies.
• Integrated semantic technologies to link guideline details with scientific literature.

Main Results:

• Created semantic models for diseases, guideline provenance, and study cohorts.
• Enabled CDSS to adapt to guideline updates and identify relevant research.
• Provided mechanisms for personalized explanations and enhanced transparency.

Conclusions:

• Enhanced existing evidence-based knowledge with ontologies and software.
• Facilitated clinician access to guideline updates, provenance, and applicable research.
• Leveraged existing biomedical ontologies and knowledge representation for explainable results.