Toward a Smart Learning Health System: An Ontology‐Based Framework

Résumé fourni par la source

ABSTRACT Background Learning health system (LHS) frameworks have been presented in multiple forms, but there is no standardized representation that captures both their core concepts and the relationships among them. This limits their practical use by health services seeking to design, implement, and evaluate LHS capabilities in changing sociotechnical environments. This study developed the Smart Learning Health System ontology (SMARTLHS), a formal semantic framework that synthesizes concepts and relationships from published LHS frameworks. Methods Relevant literature was identified through a systematic database search. SMARTLHS was developed using an iterative system mapping and ontology engineering approach, comprising five steps: (1) ontology requirements specification, (2) iterative ontology conceptualization, (3) concept and relationship comparison and formalization, (4) ontology evaluation and refinement, and (5) ontology alignment with foundational ontology, to ensure semantic consistency and interoperability. The SMARTLHS ontology was implemented using the Web Ontology Language (OWL) and developed in Protégé. Results SMARTLHS contains 447 classes and 134 object properties. Its structure is organized around four high‐level groupings: LHSs, supporting pillars, cross‐cutting themes, and common reference LHS frameworks . The supporting pillars comprise seven domains: Care delivery and organizational strategies , community and patient engagement, culture and change management, data and technology infrastructure, governance and leadership, research and education, and workforce and capacity building . The cross‐cutting themes include ethics and oversight, evaluation and methodology , and value creation and benefits . Together, these classes and relationships provide a computable semantic representation of LHS implementation. Conclusions SMARTLHS addresses key limitations of existing LHS frameworks by transforming narrative framework concepts into a formal, extensible semantic knowledge model. By explicitly representing both concepts and relationships, it provides a foundation for organizational assessment, semantic interoperability, and future AI‐enabled applications, including ontology‐guided retrieval and Retrieval‐Augmented Generation (RAG).

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.