Optimizing systematic reviews in health sciences education: a mentefact-based algorithmic framework.

INTRODUCTION

Systematic reviews in health sciences education require methodological approaches that address the pedagogical and conceptual complexity of educational interventions, which often differ from conventional clinical review models. This study addresses the research question: How can a mentefact-based algorithmic framework improve the rigor, consistency, and conceptual alignment of systematic reviews in health sciences education?

METHODOLOGY: The proposed framework comprises three linked modules: domain modeling through a mentefact, computational prioritization through rule-based filtering, semantic embeddings and clustering, and human-verified extraction and synthesis. The framework integrates the Ordinal Research Domain (ORD) and Conceptual Density and Experience Index (C-DEX) metrics to support source prioritization and conceptual relevance assessment. The workflow was structured with reference to PRISMA 2020 principles of transparency and reproducibility; however, the article reports a methodological framework with a pilot application, not a complete PRISMA-compliant systematic review.

RESULTS

The framework was piloted in the topic of artificial intelligence in radiology education, a domain in which clinical, technical, and pedagogical terminologies frequently overlap. The pilot application showed that mentefact-guided filtering can improve conceptual alignment between the educational phenomenon under review and the screening strategy used to organize the evidence.

DISCUSSION: Incorporating conceptual and pedagogical logic, rather than relying only on lexical keyword matching, may improve the interpretive relevance of evidence synthesis in health sciences education.

CONCLUSION

The framework is presented as a methodological proposal with a pilot application. Its performance requires validation in future studies using independent datasets, full reviewer-based reference standards, and formal metrics such as sensitivity, specificity, precision, recall, and inter-rater agreement.