Kosmidis Dimitrios, Simopoulos Dimitrios, Anastasopoulos Konstantinos, Koutsouki Sotiria
Department of Nursing, Democritus University of Thrace, Alexandroupolis, GRC.
Department of Medicine, Democritus University of Thrace, Alexandroupolis, GRC.
Cureus. 2025 Jul 2;17(7):e87181. doi: 10.7759/cureus.87181. eCollection 2025 Jul.
Rapid advancement of artificial intelligence (AI) and machine learning (ML) is transforming healthcare, and nursing practice is inevitably affected. Yet limited formal education leaves many nurses hesitant to integrate such tools into everyday care. This cross‑disciplinary review (i) introduces the fundamental concepts, core types and common evaluation metrices, illustrated with nursing-specific examples; (ii) catalogues the main algorithmic applications in nursing research; and (iii) describes ethical and practical challenges in their clinical use. A structured search of PubMed, Embase, Scopus, IEEE Xplore and ACM Digital Library (2015-2024) retrieved 1445 records; after de-duplication and screening against inclusion criteria (peer-reviewed, English-language studies that developed, implemented or evaluated an ML method in a clinical, community-health or educational nursing context), 61 papers were analyzed. Supervised approaches predominated, while unsupervised and semi-supervised techniques were less common. Models were evaluated mainly with accuracy, area under the receiver operating characteristic curve (AUROC), precision-recall and F1-score for classification, or mean-absolute/squared error for regression. Applications spanned eight main categories: (1) predictive risk assessment and early‑warning systems; (2) clinical decision support and diagnostic aid; (3) continuous patient monitoring; (4) workflow, staffing and operational optimizations; (5) documentation and information extraction via natural‑language processing; (6) education and competency development; and (7) other niche applications. Key barriers to wider adoption remain regulatory and ethical constraints, data quality, model transparency and engagement issues, data challenges, lack of ML-specific training in nursing curricula, and operational limitations. The utilization of ML in nursing is based on three core pillars: strong interdisciplinary collaboration, systematic integration of ML at all levels of nursing education, and a guiding framework that maps human-centered nursing interventions to interpretable ML tools. Such a foundation will enable nurses to safely leverage the results of algorithms, avoid biases and risks, and integrate ethical responsibility into technology-enhanced care.
人工智能（AI）和机器学习（ML）的快速发展正在改变医疗保健行业，护理实践也不可避免地受到影响。然而，有限的正规教育使许多护士对将此类工具融入日常护理工作犹豫不决。本跨学科综述：（i）介绍基本概念、核心类型和常见评估指标，并辅以护理领域的具体示例；（ii）梳理护理研究中的主要算法应用；（iii）描述其临床应用中的伦理和实际挑战。通过对PubMed、Embase、Scopus、IEEE Xplore和ACM数字图书馆（2015 - 2024年）进行结构化检索，共获取1445条记录；经过去重和根据纳入标准筛选（同行评审的英文研究，这些研究在临床、社区卫生或教育护理背景下开发、实施或评估了机器学习方法），对61篇论文进行了分析。监督式方法占主导地位，而无监督和半监督技术则较少见。模型评估主要采用分类的准确率、受试者工作特征曲线下面积（AUROC）、精确率-召回率和F1分数，或回归的平均绝对误差/均方误差。应用涵盖八个主要类别：（1）预测风险评估和早期预警系统；（2）临床决策支持和诊断辅助；（3）患者持续监测；（4）工作流程、人员配置和运营优化；（5）通过自然语言处理进行文档记录和信息提取；（6）教育和能力发展；（7）其他小众应用。更广泛采用的主要障碍仍然是监管和伦理限制、数据质量、模型透明度和参与问题、数据挑战、护理课程中缺乏机器学习专项培训以及操作限制。机器学习在护理中的应用基于三个核心支柱：强大的跨学科合作、在护理教育各级对机器学习进行系统整合，以及将以人为主的护理干预映射到可解释的机器学习工具的指导框架。这样的基础将使护士能够安全地利用算法结果，避免偏差和风险，并将伦理责任融入技术强化护理中。
