Search Results (1,374)

Background: Large language models (LLMs) show considerable promise for medical question answering and reasoning. Their use in medical microbiology, however, remains constrained by limited domain-specific knowledge and the risk of hallucinated outputs. Objective: To develop and evaluate Smart Medical Microbiology (SmartMM), a specialized LLM for accurate, reliable, and context-aware responses in medical microbiology. Methods: SmartMM integrates domain-adaptive continual pretraining, supervised fine-tuning (SFT), reinforcement learning from human feedback (RLHF), knowledge distillation, and retrieval-augmented generation (RAG). We constructed a high-quality microbiology corpus from textbooks, clinical guidelines, the scientific literature, case reports, and other authoritative sources. Model performance was assessed using objective examinations, subjective generation tasks, expert review, and real-world user preference evaluation. Results: SmartMM achieved accuracies of 0.897 and 0.563 on true-or-false and fill-in-the-blank questions, respectively. In subjective generation tasks, it obtained the highest ROUGE-L score (0.265) and BERTScore F1 score (0.771) among all compared models. Expert assessment showed excellent inter-rater reliability, with all ICC(C,3) values exceeding 0.970. In a user evaluation involving 20 participants and 100 real-world questions, SmartMM received the largest number of first-place rankings (33), placing it among the top-performing systems overall. Conclusions: SmartMM showed strong domain adaptability in medical microbiology knowledge organization, semantic generation, and retrieval-augmented reasoning. These findings support its potential use in educational support, infectious disease knowledge assistance, and retrieval-enhanced medical question answering. Full article

Clinical image publications represent an important educational resource across healthcare disciplines, providing concise visual demonstrations of diseases, diagnostic findings, and therapeutic outcomes. Despite their widespread use, reporting practices remain highly heterogeneous, with substantial variability in the description of clinical context, image acquisition, ethical considerations, and educational content. Existing reporting guidelines, including the CARE Guidelines, do not specifically address the unique methodological and technical aspects of image-based publications. The aim of this study was to develop the IMAGE Statement (Improving Clinical Image Reporting Standards), a reporting guideline specifically designed for clinical image publications as an extension of the CARE Guidelines. The development process was informed by recommendations from the EQUATOR Network and included a literature review, analysis of journal instructions for authors, generation of candidate reporting items, and expert consensus. The resulting IMAGE Statement comprises 15 essential reporting items covering the title, patient information, clinical context, diagnostic assessment, image acquisition, image quality, image selection, image annotation, image description, diagnostic interpretation, final diagnosis, educational message, ethics and consent, artificial intelligence use disclosure, and figure legend. The IMAGE Statement provides a structured framework intended to improve completeness, transparency, reproducibility, educational value, and ethical reporting of clinical image publications across healthcare disciplines. Full article

Insulin resistance (IR) represents one of the most pressing problems in contemporary public health. Its estimated global prevalence ranges from approximately 15.5% to over 61%, depending on the population studied, the diagnostic criteria applied, and the method used for its assessment. Despite the scale of the problem, IR remains underrecognized and lacks formal definition as a distinct disease entity, even as a growing number of clinicians and researchers worldwide describe it as such. Its asymptomatic or mildly symptomatic course allows it to remain undetected for years, during which it makes a significant contribution to the development of type 2 diabetes, cardiovascular disease (CVD) and metabolic dysfunction-associated steatotic liver disease (MASLD, formerly NAFLD), and has been increasingly linked to cellular senescence, certain cancers, neuropsychiatric disorders, and other metabolic conditions. The aim of this review was to summarize current knowledge on the pathophysiology, diagnosis, and clinical implications of insulin resistance, to discuss current challenges in its diagnosis, and to evaluate whether available scientific evidence supports its recognition as a distinct disease entity. This narrative review is based on clinical, epidemiological, and mechanistic data retrieved from PubMed and Google Scholar. Meta-analyses, systematic reviews, clinical and observational studies, clinical guidelines, and expert position statements were analyzed. Animal studies were excluded to maintain a focus on human public health implications. The diagnostic gold standard—the hyperinsulinemic-euglycemic clamp—was discussed, along with surrogate methods used in clinical practice (HOMA-IR, OGTT with insulin measurements, the TyG index, and the TG/HDL-C ratio). Factors potentially contributing to the pathogenesis of IR were examined, including hyperinsulinemia (HI), high-carbohydrate diets, inflammation, stress, and sleep disturbances, as well as conditions in which IR occurs physiologically. The findings indicate that current evidence supports the need for a clearer clinical and diagnostic framework for insulin resistance and suggest that its recognition as a distinct disease entity could facilitate earlier diagnosis, improve the standardization of clinical management, and enable earlier metabolic intervention. Given the steadily rising prevalence of metabolic disease, systemic efforts directed at the early identification and treatment of IR may be a key component of strategies aimed at reducing the population-level burden of metabolic disease and its negative consequences. Full article

Background: This scoping review maps and synthesizes the scientific evidence on the epidemiological profile and determinants of sick leave and work-related accidents in Brazil, with an emphasis on studies using data from the Brazilian Federal Medical Expert Service of the Ministry of Social Security. Methods: Following the PRISMA-ScR guidelines and the JBI Manual for Evidence Synthesis, eight databases (SciELO, PubMed/MEDLINE, Scopus, Web of Science, LILACS, CINAHL, PsycINFO, Embase) were searched for quantitative observational studies published between 2000 and 2025. Out of the 218 studies identified, 18 met the eligibility criteria. Results: Cross-sectional designs (66.7%) and ecological/time-series designs (22.2%) predominated. The findings identified musculoskeletal disorders as the leading cause of work absence; mental disorders were the third leading cause, exhibiting an upward trend; and a significant influence of institutional factors, especially the Social Security Technical Epidemiological Nexus, on the observed patterns. Determinants operate at the individual, occupational/organizational, and institutional levels. Conclusions: Empirical research on work disability in Brazil remains fragmented, providing limited capacity for causal inference. The findings support five priority areas: sectoral risk management, workplace mental health, territorial strategies for musculoskeletal disorders, sick leave trajectory policies, and data governance, all of which necessitate longitudinal and quasi-experimental studies to evaluate interventions and the impact of institutional changes. Full article

Quantum machine learning (QML) has emerged as a promising approach for analyzing the complex, high-dimensional data encountered in oncology, yet research in this area remains fragmented. This systematic literature review synthesizes current applications of QML in cancer care. Following PRISMA guidelines, six databases were searched for peer-reviewed English-language studies published between 2020 and 2026. Of the 212 records identified, 49 studies met the inclusion criteria after screening and quality assessment. The findings show that QML research is dominated by classification and detection tasks, while segmentation is beginning to emerge. Breast cancer and brain tumors are the most frequently investigated domains. Hybrid quantum-classical models, particularly quantum kernel methods, quantum neural networks, and quantum convolutional neural networks, are the predominant approaches. The main barriers to adoption are hardware limitations, including quantum noise, limited qubit availability, and reliance on simulators. Overall, QML in oncology remains in its early stages of development, with limited clinical validation, insufficient model interpretability, and little evidence of a clear quantum advantage. Future research should prioritize evaluation on real quantum hardware, larger and more diverse clinical datasets, standardized benchmarking against classical methods, and closer collaboration between computer scientists and oncology experts to facilitate clinical translation. Full article

Contact-rich manufacturing processes, such as surface cleaning, deburring, and polishing, require precise force regulation and complex trajectory tracking that are challenging to automate using conventional robot programming methods. Learning from Demonstration (LfD) offers a powerful alternative to transfer these expert skills from human operators to robotic systems. The objective of this study is to systematically map academic publications addressing LfD applications in robotic deburring and polishing between 2016 and 2026, classify the algorithmic structures, sensory modalities, and control configurations employed, and identify key industrial integration challenges. In accordance with the PRISMA 2020 guidelines, a systematic search was conducted across Scopus, Web of Science, IEEE Xplore, and Google Scholar databases. Out of the 288 initially retrieved records, duplicate removal and a two-stage screening process (Title/Abstract review, followed by full-text review) resulted in a final corpus of 24 primary studies included for qualitative synthesis. The included studies were classified into five algorithmic clusters: Dynamic Movement Primitives (DMPs) and variants (9 out of 24 studies, 38%), probabilistic and statistical models (8 out of 24 studies, 33%), deep learning and generative AI architectures (4 out of 24 studies, 17%), autonomous dynamical systems (2 out of 24 studies, 8%), and direct impedance control (1 out of 24 studies, 4%). Force/torque sensing remains the dominant modality; it was utilized exclusively in 71%—17 out of 24—of studies and in 87.5% of studies as any configuration (either as a sole modality or in multimodal setups). However, recent years have documented a trend toward multimodal perception and generative action policies (e.g., Diffusion Policies). The findings suggest that while LfD offers potential cost-reduction and flexibility benefits for small- and medium-sized enterprises (SMEs), technical barriers, such as the sim-to-real transfer gap, high-frequency impact dynamics in deburring, and the autonomous identification of local non-polishing areas (LNP areas), continue to limit widespread industrial deployment. Full article

High-altitude Andean páramo ecosystems are critical socio-ecological assets that provide key regulating and supporting services, yet they remain highly vulnerable to climate change and anthropogenic pressures. Understanding these services requires frameworks that integrate biophysical capacity with social valuation and demographic accessibility. In this study, we mapped and assessed belowground and aboveground carbon storage and species habitat provision across the altitudinal gradient (3160–4400 m.a.s.l.) of Cajas National Park (PNC), Ecuador, using a spatially explicit socio-ecological approach. Carbon stocks were modeled following The Intergovernmental Panel on Climate Change (IPCC) 2019 Refined Guidelines, combining reference soil organic carbon (SOC) stocks with aboveground biomass coefficients. Habitat provision was modeled through a multi-factor suitability index integrating Normalized Difference Vegetation Index (NDVI), elevation, water bodies, and interpolated climate surfaces. Biophysical supplies were weighted via a Delphi-based expert consultation and integrated with demographic census data (173,672 inhabitants in four adjacent parishes) to estimate actual service capture and effective benefits. Total carbon stocks in PNC were estimated at 1.49 million Mg C, with volcanic Andisols acting as the primary reservoir (97.5% of the total stock). Habitat suitability was highly heterogeneous, with critical refuge patches clustered around the park’s lacustric network and stable microclimatic zones. Socio-ecological mapping revealed that the eastern boundary of the park represents the highest hotspot of effective benefit capture, directly supplying vital water regulation and cultural services to the peri-urban populations of Sayausí and Baños. These results demonstrate the utility of integrating standard IPCC guidelines with the ECOSER protocol, offering a repeatable and transparent framework to guide spatial conservation planning and sustainable land-use zoning in vulnerable high-altitude protected areas. Full article

Background/Objectives: The treatment landscape of relapsed/refractory multiple myeloma (RRMM) has significantly evolved with the introduction of novel classes of agents and, more recently, of T-cell-redirecting therapies, including bispecific antibodies (BsAbs). It is essential to enhance and harmonize the therapeutic management of BsAbs within clinical practice. Methods: We performed a modified Delphi expert consensus study on the use of BsAbs targeting the B-Cell Maturation Antigen (BCMA) in patients with triple-class-exposed (TCE) RRMM. The study was conducted in April–November 2025 following established guidelines and best practices for defining consensus. The key phases in the use of anti-BCMA BsAbs were identified and explored. Results: Fifteen Italian hematologists with expertise in the care of TCE RRMM completed two Delphi rounds. Agreement (defined as ≥67% of panelists) was achieved on most of the topics evaluated. In particular, all panelists considered the step-up dosing phase feasible in an outpatient setting, under specific circumstances, and dosing de-escalation in responding patients to reduce the risk of adverse events. For most of them, anti-BCMA BsAb treatment is also feasible in several challenging subgroups, including frail patients (93% agreement) and those with high-risk cytogenetics (93%), extramedullary disease (93%), end-stage renal disease (86%), active plasma cell leukemia (79%), and central nervous system involvement (67%). In addition, agreement (93%) was reached on the possible sequential use of BCMA-targeting therapies, preferentially BsAbs following CAR-T, though a switch in the target antigen should primarily be considered. Conclusions: In this article, we address the main challenges related to the real-world use of anti-BCMA BsAbs in patients with TCE RRMM, offering expert recommendations to complement existing guidelines and support clinical practice. Full article

Background and Objectives: Glucagon-like peptide-1 receptor agonists (GLP-1 RAs), including semaglutide and the dual GIP/GLP-1 receptor agonist tirzepatide, have transformed the management of type 2 diabetes mellitus (T2DM) and obesity through once-weekly subcutaneous administration. While their efficacy and safety are well-established, the potential impact of administration timing—specifically, the day of the week—on clinical outcomes, adherence, and tolerability remains unexplored. Understanding whether specific days optimize therapeutic outcomes could inform personalized dosing strategies and improve long-term treatment success. Materials and Methods: We conducted a systematic literature review following PRISMA 2020 guidelines to identify studies examining the effect of weekly administration timing of semaglutide or tirzepatide on clinical outcomes in adults with T2DM or obesity. Comprehensive searches were performed across SciSpace (Deep Search, Basic Search, Full Text Search), Google Scholar, and PubMed databases through June 2026. Studies were screened using a two-stage process (abstract and full-text screening) with predefined PICO-based inclusion criteria. Data extraction focused on study design, population characteristics, administration timing, glycemic control, weight loss, adherence, and adverse events. Results: From 1471 identified records, 1000 unique papers underwent screening after deduplication and trimming. Three studies met inclusion criteria: the SUSTAIN 4 randomized controlled trial evaluating once-weekly semaglutide, a retrospective case series examining alternate-day oral semaglutide dosing, and an expert panel discussion on flexible dosing schedules. No studies directly compared different days of the week for injectable semaglutide or tirzepatide administration. Available evidence suggests that flexible dosing schedules may support adherence and tolerability without compromising glycemic control, though this suggestion derives primarily from indirect evidence and expert opinion rather than direct comparative clinical studies; direct comparative data on day-of-week effects are absent. Conclusions: Current evidence does not support a specific optimal day of the week for semaglutide or tirzepatide administration. The limited literature emphasizes flexible dosing schedules tailored to individual patient preferences and lifestyles to maximize adherence. Future prospective studies are needed to systematically evaluate whether administration timing influences clinical outcomes in real-world settings. Full article

Background: Orofacial pain (OFP) comprises a heterogeneous group of conditions with overlapping clinical manifestations that frequently complicate diagnosis and therapeutic decision-making. Diagnostic inaccuracies may contribute to delayed management, unnecessary dental procedures, and reduced quality of life. Despite the availability of standardized classifications and diagnostic criteria, their translation into routine clinical practice remains variable. Aim: This systematic review aimed to systematically identify, describe, and critically evaluate diagnostic frameworks proposed for the differential diagnosis of OFP. Diagnostic frameworks were broadly defined as structured approaches to clinical decision-making, including clinical practice guidelines (CPGs), classification systems, clinical algorithms, consensus documents, and other explicit diagnostic models. The review focused on their methodological quality, evidentiary support, clinical applicability, and operational utility in dental practice. Methods: A systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Electronic databases were searched using predefined terms related to OFP. Eligible studies included investigations assessing diagnostic protocols, clinical decision-making frameworks, or therapeutic implications in patients with OFP. Non-systematic expert opinions, editorials, and book chapters were excluded from the analysis. Data extraction and qualitative synthesis were performed independently by the reviewers. Results: A total of 184 records were identified through database searching. After duplicate removal and eligibility assessment according to the PRISMA flowchart, 12 studies were included in the qualitative synthesis. Additionally, the International Classification of Orofacial Pain (ICOP) was used as an external contextual reference framework and was not included in the PRISMA study count. The included studies demonstrated marked heterogeneity in diagnostic approaches, including theoretical, pathophysiological, clinical reasoning, classification-based, and algorithm-driven frameworks. Methodological quality was generally moderate to high. The scope and clarity domains showed the strongest performance, whereas development rigor remained limited across studies. Overall, clinically applicable frameworks frequently lacked methodological rigor. Conclusions: Current diagnostic frameworks for OFP remain methodologically heterogeneous and vary considerably in their degree of operationalization and empirical validation. Future diagnostic frameworks and CPGs should incorporate standardized terminology, reproducible diagnostic criteria, structured examination pathways, and clinically applicable recommendations to improve diagnostic consistency and facilitate implementation in routine practice. Full article

The transition to a circular bioeconomy is increasingly recognised as a critical pathway for achieving climate neutrality and resource use efficiency in Ireland. However, this transition is highly dependent on the availability of a skilled workforce capable of operating across interconnected bio-based sectors. Ireland’s National Bioeconomy Action Plan recognises the critical role that skills and knowledge will play in this transition. This study develops a comprehensive taxonomy of skills required to support the transition to a circular bioeconomy, with a specific focus on key Irish primary bioeconomy sectors (agriculture, forestry, fishery and aquaculture) and traditional industries (food processing, wood, textiles, and construction). A qualitative systematic literature review was conducted following the PRISMA 2020 guidelines. Peer-reviewed articles were retrieved from the Web of Science and Scopus databases and complemented with relevant Irish and European grey literature. A total of 96 publications were analysed using thematic analysis to identify and synthesise sector-specific and cross-sectoral competencies. The taxonomy organises circular bioeconomy skills into five overarching competence domains (systems-thinking, sustainability, circularity and bioeconomy/biotechnology, and digital skills). Findings reveal that while each sector has unique skills, a set of shared skills is consistently critical across all sectors. The identified competencies were subsequently organised into a structured taxonomy and mapped to Ireland’s National Framework of Qualifications (NFQ), establishing progressive proficiency levels from awareness (NFQ 5–6) through practitioner (NFQ 7–8) to expert (NFQ 9–10). This competency progression framework represents the principal contribution of the study, translating the taxonomy into a practical tool for curriculum development, workforce planning, professional development, and policy implementation. Full article

The treatment of osteoarthritis (OA), a leading cause of disability worldwide, is increasingly integrating nutritional strategies that appear to offer the opportunity to significantly improve the quality of care. Patients are also aware of this and often seek clarification and advice from OA professionals. However, it is still too early to formulate definitive recommendations, as the scientific evidence is still heterogeneous. This can create a gap between patients’ need for clear guidance and the necessarily prudent communication of professionals. As a result, some patients may make decisions on their own, forgoing the opportunity to receive nutritional care personalised according to their dietary habits and disease severity. This commentary presents an expert interpretation of the currently available evidence and proposes a practical framework to support clinical discussions with patients. It is not intended as a formal clinical practice guideline but aims to help OA professionals navigate conversations about the role that nutritional strategies may play in disease management. Full article

Background/Objective: Perioperative nursing care encompasses a diverse patient population with varying needs, anatomical and physiological characteristics, and differing levels of vulnerability, demanding specialized knowledge and skills. This study aims to identify key research priorities in perioperative nursing across the lifespan. Methods: A two-round e-Delphi study was conducted across three Portuguese hospitals, involving expert perioperative nurses (87 in the first round and 69 in the second). Participants rated the importance of various research topics on a 5-point Likert scale about six dimensions: (1) professional development and non-technical skills, (2) person-centered care, (3) organizational aspects and work environment, (4) patient and team safety, (5) digital innovation, and (6) patient outcomes and quality of care. Data were analyzed using mean scores, Content Validity Ratio (CVR) for initial validation, and the Coefficient of Variation (CV) to determine the level of consensus. The reporting of this study followed the DELPHISTAR (Guideline for Reporting Delphi Studies) recommendations. Results: A high degree of consensus was achieved (all CV < 0.20). The experts established a consensus on 48 components. The highest-rated priority was nurse burnout and professional satisfaction (Mean = 4.74). This was closely followed by patient safety (Mean = 4.68), team safety (Mean = 4.65), disaster response (Mean = 4.62), and emergency management (Mean = 4.58). Acute pain control and interdisciplinary communication also emerged as critical areas. Topics such as digital innovation and artificial intelligence, while valued, received lower priority scores. Conclusions: The identified priorities should direct funding toward research, education, quality improvement, and health policies. Future studies should focus on developing interventions to promote a favorable perioperative environment and optimize clinical responses to individuals’ conditions, catastrophes, and life-threatening situations across the lifespan. Full article

Introduction: Hepatitis C virus (HCV) infection often coexists with comorbidities, increasing vulnerability, complications, and adverse events. Direct-acting antivirals (DAAs) have dramatically improved HCV management, but they differ in drug–drug interaction (DDI) profiles. Sofosbuvir/velpatasvir (SOF/VEL) is associated with minimal clinically relevant interactions. Areas covered: A narrative review of the literature was conducted by searching PubMed and major international guidelines, focusing on studies published in the DAA era addressing HCV patients with major comorbidities, focusing on diabetes, metabolic syndrome, and cardiovascular disease; neuropsychiatric disorders; cancer; transplants; and use of substances or treatment with opioid agonists; and patients requiring hormone therapy including transgender patients. Expert opinion: Based on the literature and real-world data, managing polypharmacy in HCV patients with comorbidities is effective and well-tolerated, provided thorough drug review, potential DDI analysis, proactive monitoring, and coordinated multidisciplinary care are ensured. DAAs have dramatically improved the management of HCV patients; however, they have different DDI profiles that should be carefully checked. SOF/VEL has been shown to be associated with minimal clinically relevant interactions and offers a simple dosing regimen. DAA treatment is strongly advised in HCV comorbid patients not only to cure HCV but also to improve the course of comorbidities, provided that DDIs are no longer considered mere minor details. Full article

Background/Objectives: Large language models (LLMs) show promise for clinical decision support, yet their accuracy in interpreting specialized medical guidelines remains uncertain. Retrieval-augmented generation (RAG) may enhance performance by grounding responses in authoritative knowledge bases. This study aimed to compare the accuracy, comprehensiveness, and safety of RAG-enhanced versus standard LLMs for answering clinical questions derived from the German S3 guideline for oral cavity carcinoma. Methods: We conducted a prospective, single-blind benchmark study evaluating six LLMs: one RAG-enhanced model (Custom GPT with guideline access), one consensus-based model (ConsensusGPT), and four standard models (DeepSeek-V3.2, Mistral Small 3.2, Qwen3-Next-80B, GPT-OSS-120B). Fifty clinical questions covering 17 guideline domains were presented to each model three times, yielding 900 evaluations. Three expert reviewers assessed responses using 5-point Likert scales for accuracy, comprehensiveness, and clarity, under a single-blind procedure, the effectiveness of which was tested by a pre-specified manipulation check. We then ran a paired within-model experiment in which each base model was queried with and without guideline access through a transparent, openly released retrieval pipeline, and scored every response with a condition-blind automated judge alongside deterministic retrieval metrics computed from the logs. Secondary outcomes included hallucination rates and guideline citation behavior. Inter-rater reliability was assessed using intraclass correlation coefficients (ICCs). Results: In a paired within-model design that held each base model fixed, adding transparent guideline retrieval improved accuracy—significantly in the three weaker open-weight models (Mistral, Qwen3, and GPT-OSS) and directionally in the already-strong DeepSeek and GPT-5 bases. Because a pre-specified blinding check found that experts could still identify retrieval-augmented answers with 98.5% accuracy, we anchored causal interpretation on measures that do not depend on the human raters, ranked by their independence: deterministic, log-derived retrieval metrics first, and then an automated, condition-blind LLM judge, whose agreement with the experts (Spearman ρ = 0.81, 95.7% within-one agreement) establishes shared calibration rather than independence from their bias. Deterministically from the retrieval logs, citation groundedness rose from 0% to 51–89% and retrieval recall@5 was 92%. On the judge, content-level hallucination fell from 42% to 4% and accuracy rose by a pooled +0.64 points (95% CI 0.47–0.80); the accuracy gain persisted after adjustment for response length (+0.48, 95% CI 0.22–0.73), which retrieval shortened rather than lengthened. The accuracy gain was large for weaker base models and small or non-significant for already-strong ones, whereas the hallucination and auditability gains were consistent across all models. The human ratings reproduced the judge’s accuracy effect (+0.61, 95% CI 0.49–0.74), and GPT-5 run through the transparent pipeline showed no significant difference from the proprietary Custom GPT (judge accuracy 4.48 vs. 4.58). Conclusions: Guideline retrieval yields a reproducible, largely base-independent improvement in the safety and auditability of LLM answers to clinical guideline questions, with accuracy gains concentrated in weaker base models. Because retrieval-augmented answers are recognizable to experts, rigorous evaluation should rely on rater-independent measures, and residual hallucination continues to require human oversight. Full article