Sociotechnical influences on the adoption and use of AI-enabled clinical decision support systems in ophthalmology: a theory-based interview study

Background: Artificial intelligence (AI) has the potential to ease the increasing workload in ophthalmology by supporting ophthalmologists’ clinical decision-making. However, despite regulatory approvals, the adoption and use of AI-enabled clinical decision support systems (AI-CDSS) in ophthalmology remains limited. Critical obstacles that innovative healthcare technologies such as AI-CDSS face on their path to widespread clinical use are nonadoption and abandonment by their intended users, which prevent broader dissemination and real clinical benefit. This study explores how to overcome nonadoption and prevent abandonment of ophthalmic AI-CDSS by identifying ophthalmology professionals’ requirements for adoption and continued use of such tools in clinical practice.Methods: We conducted semi-structured interviews with 22 ophthalmology professionals from Germany, Austria, and Switzerland, representing a range of professional roles, clinical settings, and extent of AI-CDSS experience. To explore sociotechnical factors shaping ophthalmology professionals’ adoption decisions and their ability to derive added value from ophthalmic AI-CDSS, we conducted a qualitative content analysis combining deductive and inductive coding. The Nonadoption, Abandonment, Scale-Up, Spread, and Sustainability (NASSS) framework deductively guided the development of higher-level code categories, representing seven sociotechnical domains relevant for the implementation of healthcare technologies. These were further refined into subcategories through inductive coding of the interview material.Results: Most participants expressed general openness to ophthalmic AI-CDSS. However, actual adoption decisions and the ability to derive clinical value from these tools were shaped not just by individual attitudes but also by a range of other sociotechnical influences. Specifically, we inductively identified 29 code categories, representing sociotechnical influences and requirements from all seven NASSS domains, including technological, user, organizational, and societal aspects. Our findings also suggest that while many sociotechnical influences and challenges are shared between AI-based and traditional healthcare technologies, a key distinction of (ophthalmic) AI-CDSS lies in the users’ psychological appraisal of such tools. Conclusions: Our findings highlight the complex and context-specific nature of integrating AI-CDSS into ophthalmic practice. This study also informs AI and healthcare researchers on the applicability of the NASSS framework for studying AI implementation and provides actionable insights for AI developers and implementers aiming to address user needs more effectively.