JMIR Publications

ABSTRACT

Background:

Widespread and sustained uptake of artificial intelligence (AI)-based clinical decision support systems (CDSS) in real-world healthcare environments is uncommon, despite their potential to improve patient care and reduce clinician burnout. Although previous studies have studied determinants to implementing AI-based CDSS, their perspectives focused on theoretical or pre-implementation considerations.

Objective:

The objectives of this scoping review were to (1) map and synthesize barriers and facilitators to implementing AI-based CDSS in real-world healthcare scenarios and (2) draw on this knowledge to inform future implementation strategies.

Methods:

Five electronic databases were searched for studies describing the real-world implementation of AI-based CDSS in any healthcare setting. Barriers and facilitators were extracted from eligible studies by two independent reviewers and described using the Consolidated Framework for Implementation Research to aid in the formulation of future implementation strategies.

Results:

Of the 14,301 articles screened, only 13 were eligible for inclusion in the study. Commonly identified barriers included challenges related to algorithm interpretability, data management and quality, mismatches between end-user needs and system functions, and end-users’ capability and motivation to use the CDSS. Conversely, early and continued assessments of end-user needs, peer endorsement and engagement, and the availability of robust evidence backing CDSS use were frequently identified as facilitators.

Conclusions:

Understanding how determinants influence the successful implementation and sustained use of AI-based CDSS in real-world scenarios is a fundamental step in developing robust and comprehensive strategies to realize their potential benefit. The determinants identified in this review offer foundational insights that can be used as a basis for future implementation strategies.