Scoping review of the methods used to map clinical expertise and knowledge in the creation of knowledge-based clinical decision support systems
Katie Bounsall, Cen Cong, John Downey, Precious Onyeachu, Victoria Haunton, Victoria Allgar, Camille B CarrollBackground
Digital health technologies are rapidly increasing within healthcare services; however, adoption remains limited. Clinical decision support systems (CDSSs) are digital software designed to support decision-making. Reporting practices for CDSS development remain inconsistent and methodological guidance is limited.
Objectives
To identify and synthesise the methods used to capture and map clinical knowledge for the development of knowledge-based CDSSs.
Eligibility criteria
English-language studies (2012–2024) describing the development of knowledge-based CDSSs and reporting of the methods used to construct their knowledge bases. Eligible study designs included qualitative, quantitative and mixed-methods research. Commentaries, editorials and conference abstracts without full texts were excluded.
Sources of evidence
PubMed, IEEE Xplore, Web of Science and Embase were searched. No protocol was published.
Charting methods
Data were extracted on study characteristics, clinical domain and approaches used to capture and structure knowledge for CDSS development. Methods were synthesised descriptively.
Results
Of 10 578 records screened, 67 studies met the inclusion criteria. Studies spanned 24 countries and 46 clinical areas. Five main approaches to knowledge-base creation were identified: consultation with domain experts, use of clinical practice guidelines, literature reviews, ontologies and historical clinical data. Domain-expert input was the most common method (65%; 44/67), although reporting was inconsistent, with 61% (27/44) studies not specifying the number of experts involved and 45% (20/44) not reporting their professional backgrounds. Across studies, substantial variation and gaps were observed in the documentation of knowledge-capture processes.
Conclusions
Methods for constructing knowledge bases in CDSSs are highly heterogeneous and often poorly reported. Greater standardisation—potentially through dedicated reporting guidelines or methodological frameworks may strengthen transparency, reproducibility and the design quality of future CDSSs.