Artificial intelligence in nursing support for patients: A rapid review.

Abstract

This study aimed to gain insights into the potential of using evidence from artificial intelligence (AI) in nursing support for clinical practice and decision-making and its implications for future studies. This was a rapid review of the literature. The PubMed, CINAHL, and CENTRAL in the Cochrane Library databases were searched for randomized controlled trials (RCTs) on nursing support using AI for patient care, published between 2010 and 2024. The included studies were assessed for quality appraisal using the mixed methods appraisal tool, version 2018, and data extracted from each study were synthesized narratively considering the review questions. After removing duplicates, 5,176 studies were identified, resulting in the inclusion of 20 studies. The 20 studies were published between 2012 and 2024, with five studies focusing on "providing information and advice to patients using AI," seven studies on "clinical decision-making support for nurses using predictive algorithms," and eight studies on "psychosocial support using AI-equipped social robots." In the qualitative appraisal of these studies, the proportion of studies that met the quality criteria for each item ranged from 35% to 85%. In this review, we identified 20 RCTs that have been published so far on AI in nursing support, and currently, research is shifting from the development and testing phase to the implementation and effectiveness verification phase. However, many studies were found to be at risk of bias, suggesting caution when applying the evidence to clinical practice. Moving forward, there is a pressing need for the accumulation of high-quality RCTs and evidence integration through meta-analyses. This study explored the potential of AI in nursing support, highlighting 20 published RCTs. Research in this field is transitioning from development and testing to implementation and effectiveness verification.