This study introduces a novel methodology for enhancing intelligent tutoring systems (ITS) through the integration of generative artificial intelligence (GenAI) and specialized AI agents. We present a proof of concept (PoC) demo that implements a dual-layer GenAI validation approach that utilizes multiple large language models to ensure the reliability and pedagogical integrity of the AI-generated content. The system features role-specific AI agents, a GenAI-powered scoring mechanism, and an AI mentor that provides periodic guidance. This approach demonstrates capabilities in dynamic scenario generation and real-time adaptability while addressing key challenges in AI-driven education, such as personalization, scalability, and domain-specific knowledge integration. Although exemplified here through a case study in healthcare root cause analysis, the methodology is designed for broad applicability across various fields. Our findings suggest that this approach has significant potential for advancing adaptive learning and personalized instruction while raising important considerations regarding ethical AI application in education. This work provides a foundation for further research into the efficacy and impact of GenAI-enhanced ITS on learning outcomes and instructional design across diverse educational domains.

Objective: This study synthesises the effectiveness and applicability of trigger tools for detecting adverse events (AEs) across various healthcare settings. Design: This study used an umbrella review to consolidate findings from existing systematic reviews, assess the quality of evidence and identify gaps in current knowledge. Methods: A comprehensive search was conducted across SCOPUS, Web of Science and PubMed, and included systematic reviews and meta-analyses from 2009 to 2024 focusing on trigger tools used in healthcare settings. Data extraction and quality appraisal followed Joanna Briggs Institute guidelines. Narrative synthesis was employed owing to the heterogeneity among studies. Nine systematic reviews were included. Results: Nine systematic reviews were analysed. Results showed substantial variation in trigger tool performance. Detection rates ranged from 0.8% to 66% across healthcare settings and populations. Preventability rates varied widely from 7% to 94.4%. Automated detection methods demonstrated inconsistent results with AE prevalence, ranging from 0.1% to 29.2%. The studies within the reviews exhibited significant variability in the types of trigger tools analysed and the methodologies used, affecting detection outcomes. The quality of the included studies varied, with inconsistent definitions of AEs and differences in study design limiting the generalisability of the findings. Conclusions: Trigger tools vary significantly in effectiveness, influenced by healthcare context and tool design. Automated methods need refinement. Standardised methodologies and context-specific tools are essential to improve patient safety. Future research should focus on these areas to enhance the accuracy and applicability of trigger tools in different healthcare settings. PROSPERO registration number: CRD42024581456.

Osteoporosis, a prevalent skeletal disorder, necessitates accurate and accessible diagnostic tools for effective disease management and fracture prevention. While dual-energy X-ray absorptiometry (DXA) remains the clinical standard for bone mineral density (BMD) assessment, its limitations, including ionizing radiation exposure and susceptibility to artifacts, underscore the need for alternative technologies. Ultrasound-based methods have emerged as promising radiation-free alternatives, with radiofrequency echographic multi-spectrometry (REMS) representing a significant advancement in axial skeleton assessment, specifically at the lumbar spine and proximal femur. REMS analyzes unfiltered radiofrequency ultrasound signals, providing not only BMD estimates but also a novel fragility score (FS), which reflects bone quality and microarchitectural integrity. This review critically evaluates the underlying principles, diagnostic performance, and clinical applications of REMS. It compares REMS with DXA, quantitative computed tomography (QCT), and trabecular bone score (TBS), highlighting REMS’s potential advantages in artifact-prone scenarios and specific populations, including children and patients with secondary osteoporosis. The clinical utility of REMS in fracture risk prediction and therapy monitoring is explored alongside its operational precision, cost-effectiveness, and portability. In addition, the integration of artificial intelligence (AI) within REMS software has enhanced its capacity for artifact exclusion and automated spectral interpretation, improving usability and reproducibility. Current limitations, such as the need for broader validation and guideline inclusion, are identified, and future research directions are proposed. These include multicenter validation studies, development of pediatric and secondary osteoporosis reference models, and deeper evaluation of AI-driven enhancements. REMS offers a compelling, non-ionizing alternative for axial bone health assessment and may significantly advance the diagnostic landscape for osteoporosis care.

Quality improvement processes seek to establish effective practices as everyday habits. In other words, they aim to make evidence-based practices the norm. There are several barriers to consider. First, there is a tendency to conflate measured change with embedded practice. For example, hand hygiene initiatives benefit from a Hawthorne effect. People may change their behavior when they know they are being observed. And some observation tactics and settings are more noticeable than others. Second, measurements are often made at the project’s peak of attention rather than during routine practice after the intervention. Factors that may distinguish initial engagement from sustained integration include attenuation over time, variation between implementation sites, and variation by observation conditions. Effective quality improvement projects will anticipate and measure these factors, report the conditions under which data were collected, and continue evaluation beyond the active project phase. Sustained change also depends on organizational conditions that support routine practice. Improvement is not change until it no longer depends on being watched.