Readiness Assessment for AI in Nursing Care Projects: Multimethods Study.

BACKGROUND: Integrating artificial intelligence (AI) systems into nursing care often encounters obstacles stemming from unmet requirements and insufficient engagement with well-documented sociotechnical pitfalls. Readiness models offer a systematic way to evaluate project preparedness and to build the capabilities needed for successful artificial intelligence in nursing care (AINC) research, development, and implementation. As of yet, an evidence-based AI readiness assessment prioritizing AINC projects and accounting for their diversity in care settings is missing. OBJECTIVE: This study aimed to develop a comprehensive artificial intelligence nursing care readiness assessment (AINCRA) to support planning, execution, and evaluation of AINC projects. METHODS: In a sequential exploratory multimethods bottom-up approach to maturity model development, key AI readiness dimensions and attributes were identified to develop a pilot readiness assessment. The pilot version was grounded on insights from an expert workshop (n=21) and expert interviews (n=14), an online survey (n=53), a rapid review (n=292), and a nominal group consensus process. A systematic literature review (n=7) further triangulated AI readiness attributes. Finally, a think-aloud interview study and focus group discussions involving experts (n=18) from nursing practice, nursing science, and AI research and development who had conducted AINC projects prior to data collection validated the attributes. RESULTS: The resulting AINCRA encompasses 5 core dimensions: regulatory, processual, technical, social, ethical, and community building requirements and aspects. Including 69 attributes and capabilities of AI nursing care readiness, the core dimensions reflect key areas of action where AINC project stakeholders can influence project outcomes. Clinical partners can assess their organization's maturity level in relation to the implementation of AI. An assessment of each dimension and its attributes across 5 maturity levels allows reflecting on and proactively shaping individual project approaches. Overall, experts regarded AINCRA as a useful instrument for the development, management, and evaluation of AINC projects while emphasizing that established principles of good practice in project and data management should not be neglected when using AINCRA as a project management tool. CONCLUSIONS: AINCRA enables practitioners from AI research and development, clinical partners, and nursing and health scientists to plan, evaluate, and enhance AI projects across their lifecycle, thereby supporting effective AI integration in nursing care. While AINCRA was developed within the European and German legal framework for AI in health care settings, respective attributes can be adapted to international requirements.