Mapping Artificial Intelligence-Mediated Nursing Processes in Intensive Care Units for Critically Ill Oncology Patients: A Scoping Review Protocol

This scoping review protocol maps the current state of literature on Artificial Intelligence (AI)-mediated Nursing Processes (Sistematização da Assistência de Enfermagem – SAE) in Intensive Care Units (ICUs) for critically ill oncology patients. The review focuses specifically on how AI technologies — including Large Language Models (LLMs) and Clinical Decision Support Systems (CDSS) — integrate with standardized nursing taxonomies (NANDA-I, NOC, and NIC), and on the validation metrics used to assess these systems, namely diagnostic accuracy, Content Validity Index (CVI), and Kappa coefficient. The purpose of this review is twofold. First, it addresses a genuine gap in the literature: preliminary searches across PubMed/MEDLINE, Scopus, ScienceDirect, and OSF Registries confirmed that no existing scoping review or registered protocol simultaneously covers the intersection of AI-mediated Nursing Process, the ICU context, and the critically ill oncology population evaluated through these specific validation metrics. Related reviews exist for AI in nursing decision-making broadly, AI in critical care nursing, AI in oncology nursing, and computerized decision-support systems built on NNN taxonomies — but none combine all three elements together. Second, the review responds to a practical need in critical care nursing: as AI tools increasingly assist nurses in formulating diagnoses, predicting outcomes, and selecting interventions in high-acuity settings, the validity, safety, and clinical reliability of these AI-generated outputs must be systematically understood before broader adoption. Following the JBI methodology for scoping reviews (Peters et al., 2020/2024) and reported according to the PRISMA-ScR checklist (Tricco et al., 2018), the review will search PubMed/MEDLINE, Scopus, Web of Science, CINAHL, LILACS, Google Scholar, and ProQuest Dissertations and Theses for quantitative, qualitative, and mixed-methods studies — including grey literature — addressing AI-driven nursing systems in adult ICUs, with a specific focus on oncology patients. Two independent reviewers will conduct title/abstract and full-text screening, followed by structured data extraction covering study metadata, population/setting, AI technology type, taxonomy use, validation metrics, and key findings. Expected outcomes include: a comprehensive map of how AI technologies are currently being integrated into the Nursing Process for critically ill oncology patients; a characterization of the methodological approaches used to validate these AI systems (particularly diagnostic accuracy, CVI, and Kappa agreement); identification of technological and methodological gaps in the field; and a narrative synthesis highlighting clinical benefits, ethical considerations, and implementation challenges. These findings are intended to guide future empirical research and inform the responsible development of AI-assisted nursing tools in high-acuity oncology critical care settings.