سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / اولین همایش بین المللی تحول خدمات سلامت: تلفیق مراقبت، درمان، بازتوانی و امیدبخشی

Large Language Models in Nursing Practice: Opportunities, Risks, and Future Directions (Narrative Review)

نویسندگان :

Melika Hosseini¹ Mansoureh Ashghali Farahani² Anita Pouzesh³

1- دانشگاه علوم پزشکی ایران 2- دانشگاه علوم پزشکی ایران 3- دانشگاه علوم پزشکی ایران

کلمات کلیدی :

large language models،generative artificial intelligence،nursing practice،clinical documentation،nursing education،health informatics

چکیده :

Background: Large language models (LLMs) are entering nursing education, documentation, care planning, communication, and clinical decision support. Their ability to generate and transform clinical language may reduce administrative workload and improve access to tailored information, but their use raises concerns about accuracy, bias, privacy, explainability, and professional accountability [1-5]. Method: This narrative review synthesized recent scientific literature published between 2023 and 2026. A total of 5 relevant articles were included in the narrative synthesis. Findings were organized thematically around clinical utility, education, documentation, patient safety, ethics, and implementation. Finding: LLMs may support care-plan development, handover and documentation, patient education, faculty-supported learning, and retrieval of locally grounded clinical knowledge. Emerging development, usability, and conceptual evidence suggests potential gains in workflow efficiency and information standardization; however, outputs remain probabilistic and may contain hallucinations, omissions, bias, or unsafe recommendations. Responsible adoption requires human verification, transparent disclosure of AI involvement, privacy-preserving data practices, competency-based education, local validation, and clear accountability. Evidence regarding patient outcomes, equity, long-term workload, and nurse-patient relationships remains limited [1-5]. Conclusion: LLMs should be implemented as supervised augmentation tools rather than autonomous decision-makers within nurse-led, evidence-based workflows. Prospective, multisite evaluations and co-design with nurses, patients, educators, informaticians, and regulators are priorities. Keywords: large language models; generative artificial intelligence; nursing practice; clinical documentation; nursing education; health informatics