Factors influencing healthcare providers’ adoption of artificial intelligence-based clinical decision-support systems: a systematic review using the CFIR framework

Abstract

Artificial intelligence-based clinical decision-support systems (AI–CDSS) are increasingly used in healthcare, but their adoption by healthcare providers (HCPs) remains challenging. Reported determinants span technological, organizational, individual, and implementation-related levels. To systematically review factors influencing HCPs’ adoption of AI–CDSS and organize them using the Consolidated Framework for Implementation Research (CFIR). We conducted a mixed systematic review of studies retrieved from Web of Science, PubMed, EMBASE, Scopus, and IEEE through December 2024. Eligible studies examined HCP acceptance or use of AI–CDSS in actual or simulated deployment. Qualitative findings were synthesized thematically using an operational adaptation of the original CFIR, and heterogeneous quantitative findings were synthesized narratively. A total of 11,015 records were retrieved, and 20 studies were included. Eleven studies assessed actual deployments, eight assessed simulated deployments, and one included both contexts. Thirteen studies evaluated AI–CDSS exclusively in hospital settings, four evaluated AI–CDSS exclusively in primary care settings, and three included both settings. Most studies used qualitative ( n =10) or mixed-methods designs ( n =7). The CFIR-guided synthesis organized determinants across five domains and summarized a small number of relationships reported between selected themes. Evidence concerning Process and Outer Setting determinants remained limited. A comprehensive approach that engages HCPs in AI–CDSS design and development and provides targeted training on HCPs’ practical skills and operational principles of AI–CDSS is more likely to promote acceptance and use of AI–CDSS in real clinical settings. Trial Registration The protocol was registered with PROSPERO (CRD420250655501).