Explore the vast range of titles available.

This study aimed to provide an in-depth understanding of English as a Foreign Language (EFL) students' perceptions concerning both the benefits and limitations of Artificial Intelligence (AI) in the context of critical thinking. Utilizing a qualitative research design that focuses on case studies and employs semi-structured interviews, seven master's degree students from two different Indonesian universities were purposively selected for the sample. The findings revealed a complex view of critical thinking that involves questioning norms, analyzing context, and evaluating evidence. Students acknowledged AI's utility in enriching various facets of critical thinking, such as academic research and theory scrutiny. However, concerns were also raised about AI's limitations, including lack of personalization, potential for echo chambers, and difficulties in nuanced understanding. The study concludes that AI can be an asset in the development of critical thinking skills, but with caveats that require careful management. A balanced approach that capitalizes on AI's strengths while being aware of its limitations is necessary for cultivating robust critical thinking abilities among EFL students. Limitations of the study include its reliance on self-reported data, which may introduce biases, and the heterogeneity in the participants' backgrounds, affecting generalizability. Future research may consider more objective measures such as observations or psychometric tests, and investigate pedagogical methods for integrating critical thinking and AI applications effectively.

Aim To synthesize recent research on artificial intelligence (AI) in intensive care unit (ICU) nursing from 2020 to 2025, highlight trends, and outline integration challenges. Methods A narrative synthesis approach was used, reviewing English-language studies from PubMed, Web of Science, Scopus, and IEEE Xplore. From 4138 articles, 37 studies were included. Results Evidence was international with strong contributions from Asia and North America. Most studies were retrospective and drew on large ICU databases such as MIMIC-III/IV and eICU. Methods were dominated by machine learning, with limited but growing deep learning. Applications clustered around early warning and risk prediction, with additional work on nursing decision support and workload or documentation support. Reported discrimination frequently exceeded AUC 0.80, while calibration, external validation, and human factors evaluation were less often described. Conclusion Artificial intelligence shows promise for earlier risk recognition, decision support, and workflow enablement in ICU nursing. Priorities include multicenter prospective evaluation, external validation with calibration, electronic health record-embedded implementation, and nurse codesign to ensure safe, useful, and generalizable tools. Implications for clinical practice Thoughtfully integrated AI can support timely decisions and reduce documentation burden when paired with real-time validation and nurse-led workflow adaptation.

Large language models (LLMs) have advanced rapidly. However, concerns remain regarding their reliability in clinical settings due to the inherent issues of hallucinations and inadequate referencing. We evaluated six current LLMs: GPT-4.1 (GPT), o3, Gemini-2.5-Pro-Preview-0506 (Gemini), Grok-3 (Grok), Qwen3-235B-A22B (Qwen3), and Claude Sonnet 4 (Claude), as well as two technologies that extend LLM capabilities using external knowledge bases: retrieval-augmented generation (RAG) and Model Context Protocol (MCP). Each model was evaluated using 50 questions selected from a 132-question pool developed based on the Chinese Medical Association guideline for clinical diagnosis and treatment of lung cancer (2024 Edition). Three models-Qwen, GPT, and Grok-were further analyzed to assess performance changes with RAG and MCP integration. All responses were independently reviewed by two qualitative evaluators. Overall, o3 achieved the highest accuracy (50%), followed by GPT (48%) and Gemini (48%), then Grok (44%), Qwen (40%), and Claude (36%). However, implementing RAG (LLM-RAG) or MCP (LLM-MCP) significantly improved accuracy, with statistical differences observed between baseline LLMs and their RAG- or MCP-enhanced counterparts. Lexical richness and semantic noise both diminished, whereas the semantic clarity and accuracy of verbs, noun-verb combinations, and content words improved. The six latest LLMs performed similarly on lung cancer-related questions. The integration of RAG or MCP significantly enhanced accuracy while simplifying sentence structure, focusing more on the main topics, and using more accurate vocabulary.

This study evaluated the performance of five Large Language Models based on actual cases to provide guidance for selecting appropriate models for clinical decision-making. This study aimed to assess the performance of large language models (LLMs) in clinical decision-making for internal medicine and to provide evidence-based guidance for model selection in clinical practice. We conducted a retrospective cross-sectional study with 405 cases across nine subspecialties: cardiovascular, respiratory, gastroenterology, nephrology, rheumatology, endocrinology, neurology, hematology, and infectious diseases. Two senior clinicians evaluated outputs on five dimensions: diagnosis, diagnostic criteria, differential diagnosis, examinations, and treatment. Statistical analyses were performed via the Kruskal‒Wallis tests and Pairwise comparisons were performed by Dunn's test with p-value adjusted by BH procedure. Overall, significant performance differences were observed among models (  = 0.001). All models performed worst in respiratory (  < 0.05). Gemini significantly outperformed others in differential diagnosis-the weakest area across all models (  < 0.05). Claude scored significantly lower than other LLMs in Card and Heme (  < 0.05). Subgroup analysis indicated that the most pronounced performance disparities were observed in the Card (  < 0.05). GPT, O1, and Gemini demonstrated superior performance in clinical decision-making for internal medicine among all LLMs, whereas Claude showed the poorest performance. All LLMs demonstrated deficiencies in differential diagnosis and poor management for respiratory diseases. The complexity of subspecialty might be a performance differentiator for LLMs and O1 might have potential suitability for complex subspecialties like cardiology.

Background The integration of artificial intelligence (AI) into dentistry has potential to improve diagnostic accuracy and enable more personalized treatment planning. However, the practical and legal aspects of obtaining informed consent in this context remain unresolved. Objectives The present article aimed to unify current positions found in the ethics literature with current legal standards in order to develop clinically applicable decision flowchart and checklist for patient consent in AI-assisted dentistry. Methods A qualitative thematic synthesis of 50 publications addressing AI, ethics, and informed consent in dentistry was conducted using Braun and Clarke's six-phase framework. Insights were translated into a structured consent checklist and a decision-making flowchart, which were then evaluated through a pilot survey of practicing dentists in Serbia who use AI tools. Results Twelve dentists completed the survey. Most rated the checklist and flowchart as clear, practical, and feasible for clinical application. Over 90% agreed that standardized consent improves patient communication and trust when AI is explained as being under clinician supervision. Conclusions The proposed consent checklist and flowchart provide a clinically applicable framework that strengthens patient autonomy, supports ethical and transparent AI use in dentistry, and offers guidance for practitioners navigating emerging regulatory requirements.

Detecting drowsiness is crucial for improving traffic safety and preventing fatigue-related accidents. This paper aims to develop an advanced real-time drowsiness detection system using deep learning algorithms. For this purpose, we utilized an eye image dataset from the MRL Eye Dataset and performed extensive feature engineering and preprocessing to prepare the data for analysis. An algorithm has been proposed to classify eye states as open or closed using Transfer Learning based on the MobileNet architecture. Using a balanced dataset, the model was trained to distinguish between open and closed eyes accurately. The experimental results demonstrate the effectiveness of our approach, achieving 90% accuracy, 100% precision, 83.3% recall, and an F1-score of 90.9%. To further validate the system, we integrated the trained model into a real-time camera application that monitors the eye conditions of drivers. The application analyzes real-time video streams, detects faces and eyes, and uses instances of closed eyelids to predict signs of drowsiness. The efficiency of the model is evaluated using metrics such as accuracy, precision, recall, and F1-score. The results indicate that our approach accurately identifies fatigue indicators, presenting a viable solution for real-time drowsiness monitoring to help prevent accidents caused by exhaustion. Future studies will explore incorporating additional physiological information and applying advanced deep-learning techniques to enhance detection accuracy and system robustness.

Background As artificial intelligence (AI) becomes increasingly integrated into healthcare and nursing education, existing competency standards and training systems for clinical nursing educators are no longer sufficient to meet the demands of technology-enabled teaching. This study aimed to explore the lived experiences and perceived competency demands and training needs of clinical nursing educators in the AI era, and to provide empirical evidence to inform competency framework development and training-system optimization. Methods A descriptive phenomenological qualitative study was conducted in four tertiary hospitals in Chengdu, China, between July and September 2025. Fourteen clinical nursing educators were recruited using purposive sampling. Data were collected through semi-structured interviews and analyzed using Colaizzi’s seven-step method. Results Five themes and 17 subthemes were identified. The core competency framework comprised four key domains: professional foundation and specialized technical skills, communication and humanistic care, AI tool application, and research translation. Training-system optimization involved four interrelated areas: systematic training design, incentive and support mechanisms, specialty-based and tiered teaching design, and the educational value and practical challenges of integrating AI into clinical teaching. These findings suggest that clinical nursing educators require both foundational teaching competencies and new AI-related capabilities. Conclusions Clinical nursing educators in the AI era require a competency-based development pathway that integrates pedagogical, clinical, technological, and research capabilities. Hospitals and nursing education institutions should establish tiered training programs, practice-oriented assessments, and supportive organizational policies to strengthen educator preparedness and facilitate the intelligent transformation of clinical nursing education. Trial registration This is a qualitative study, not a health care intervention trial, so no trial registration is applicable.

The rapid integration of artificial intelligence (AI) into organizational workflows is re-shaping traditional patterns of interaction between leaders and employees. Grounded in power dependence theory, this study investigates how employees' voluntary dependence on AI influences leader-subordinate power relations and, consequently, influences employees' voice behavior. We propose that employees' dependence on AI can increase their perceived power when interacting with leaders, which subsequently enhances their willingness to offer constructive suggestions or question established practices. Furthermore, we propose that the extent to which leadership tasks can be substituted by AI plays a moderating role in this process. Coaching leadership, characterized by its emphasis on guiding task performance and developing employee skills, may be particularly sensitive to such substitution. Using two experimental studies and two survey investigations, we provide evidence that employees' AI dependence is positively associated with voice behavior through heightened perceptions of personal power, and that this relationship is strengthened under high levels of coaching leadership. These findings advance leadership theory by explicating how AI adoption alters foundational power structures in the workplace and by identifying a novel, power-based pathway linking AI use to proactive employee behaviors. The study contributes to emerging discussions on effective leadership in technologically augmented organizations and offers empirical insights into how leaders can adapt their roles and behaviors in the new era of AI-driven work.