Nurse education in practice

1873-5223932026MayNurse education in practiceNurse Educ PractGenerative AI and intelligent tutoring systems in nursing education: A systematic review of impacts on higher-order thinking skills and clinical competency.10484710484710.1016/j.nepr.2026.104847S1471-5953(26)00149-6This review synthesizes the impact of Artificial Intelligence (AI) on nursing students' clinical competency, Higher-Order Thinking Skills (HOTS) and educational outcomes to differentiate between procedural efficiency and deep cognitive retention.Generative AI and Intelligent Tutoring Systems (ITS) offer personalized scaffolding. However, their comparative efficacy against conventional pedagogical approaches, such as face-to-face lectures and standard non-AI simulations, remains controversial, particularly regarding the potential trade-off between skill acquisition and cognitive depth.A systematic literature review guided by PRISMA 2020 guidelines.A comprehensive search across seven databases (PubMed, EBSCOhost, Scopus, ScienceDirect, ProQuest, IEEE Xplore, JSTOR) from 2015 to 2025 identified 14 eligible studies (7 RCTs, 7 Quasi-experiments) which were synthesized across East Asia, Europe and Africa. Quality was appraised using JBI tools, with the protocol registered in PROSPERO.Synthesis of 14 studies (n = 1107) reveals a critical divergence. AI interventions significantly improved psychomotor application, procedural skills and communication mechanics compared with traditional instructor-led teaching methods. However, findings on HOTS and retention were mixed; while GenAI enhanced inquiry-based problem solving, it occasionally compromised critical reflection and long-term knowledge retention compared with human-led instruction. Furthermore, a "gap in algorithmic empathy" was evident, with AI failing to effectively enhance cultural awareness.AI functions as a robust "pedagogical scaffold" for procedural simulation but risks inducing "cognitive offloading" if used passively. Educators must adopt a Hybrid-Scaffolded Model: leveraging AI for iterative drills while prioritizing human facilitation for deep cognitive consolidation, ethics and cultural safety.Copyright © 2026 Elsevier Ltd. All rights reserved.NasirunHafizsHFaculty of Nursing, Universitas Indonesia, Depok, West Java, Indonesia. Electronic address: hafizs.nasirun@ui.ac.id.MulyonoSigitSDepartment of Community Nursing, Faculty of Nursing, Universitas Indonesia, Depok, West Java, Indonesia. Electronic address: sigit@ui.ac.id.HapsariAyu Widowati DwiAWDFaculty of Mathematics and Natural Sciences, Universitas Indonesia, Depok, West Java, Indonesia. Electronic address: Ayu.widowati@ui.ac.id.WiradinataHendri HardiHHFaculty of Nursing, Universitas Indonesia, Depok, West Java, Indonesia. Electronic address: hendri.hardi@ui.ac.id.RatuMustikaMFaculty of Nursing, Universitas Indonesia, Depok, West Java, Indonesia. Electronic address: mustika.ratu@ui.ac.id.engJournal ArticleSystematic Review20260428

2306-53541332026Mar20Bioengineering (Basel, Switzerland)Bioengineering (Basel)Applications of Large Language Models in Medical Research: From Systematic Reviews to Clinical Studies.36510.3390/bioengineering13030365Background: Large Language Models (LLMs) are reshaping medical research workflows. Objective: This narrative review synthesizes evidence on LLM applications across systematic reviews, scientific writing, and clinical research. Methods: We reviewed literature from 2023-2025 examining LLM applications in medical research, identified through PubMed, Scopus, Web of Science, arXiv, medRxiv, and Google Scholar. Studies reporting empirical findings, methodological evaluations, or systematic analyses of LLM applications were included; editorials and commentaries without empirical data were excluded. Results: In systematic reviews, LLMs achieve 80-94% data extraction accuracy and 40% reduction in screening workload, but show only slight-to-moderate agreement (κ = 0.16-0.43) in risk-of-bias assessment. In scientific writing, hallucination rates of 47-55% for fabricated references and over 90% prevalence of demographic bias require rigorous verification. For clinical research, LLMs assist with statistical coding and protocol development but require human validation. Critically, excessive reliance on automated tools may cause cognitive offloading that compromises analytical capabilities. Conclusions: LLMs are powerful but unstable tools requiring constant verification. Success depends on maintaining human-in-the-loop approaches that preserve critical thinking while leveraging AI efficiency.GongEun JeongEJ0000-0003-3996-3472Department of Internal Medicine, Hallym University College of Medicine, Chuncheon 24252, Republic of Korea.Institute for Liver and Digestive Diseases, Hallym University, Chuncheon 24252, Republic of Korea.Institute of New Frontier Research, Hallym University College of Medicine, Chuncheon 24252, Republic of Korea.BangChang SeokCS0000-0003-4908-5431Department of Internal Medicine, Hallym University College of Medicine, Chuncheon 24252, Republic of Korea.Institute for Liver and Digestive Diseases, Hallym University, Chuncheon 24252, Republic of Korea.Institute of New Frontier Research, Hallym University College of Medicine, Chuncheon 24252, Republic of Korea.ShinYong SeokYSDepartment of Internal Medicine, Hallym University College of Medicine, Chuncheon 24252, Republic of Korea.engJournal ArticleReview20260320