Search Results (166)

International shipping decarbonization has expanded beyond technical and operational regulation to include contested market-based bargaining. The proposed International Maritime Organization (IMO) Net-Zero Framework (NZF) combines a marine fuel standard with an economic compliance mechanism. Drawing on IMO submissions, voting records, official statements, and domestic and sectoral materials, this two-level game analysis examines why support sufficient to approve the draft for circulation in April 2025 did not translate into timely adoption later that year. The evidence suggests that the initial agreement space was not durable enough to support adoption on the planned timetable. At Level I, bargaining involved three linked dimensions: rule-making authority, responsibility allocation, and transition pathway design. At Level II, domestic, regional, and sectoral constraints included regulatory credibility and investment signals for the European Union; sovereignty and cost concerns for the United States; trade exposure and disproportionate impacts for trade-exposed emerging economies; ambition and revenue-supported transition for small island developing States; and fuel availability, fleet competitiveness, and legal certainty for energy exporters and open registries. A coalition sufficient at one procedural stage may therefore not remain sufficient at the next. A more durable agreement space would likely require an IMO-centered regulatory core, targeted enabling support, an ambition floor with compliance flexibility, and sequenced implementation and review. Full article

Context: The Agile methodology has been prevalent in the software industry for more than two decades, marking a shift from plan-driven to market-driven approaches and introducing various challenges. While the literature identifies numerous challenges in Agile development, little attention has been given to their ranking and prioritization, which are critical for effective project management and decision making. This study fills this gap by combining empirical evidence from the literature and practitioners. Objectives: This study aims to identify and hierarchically prioritize the most recent challenges faced by Agile practitioners during product development. To achieve this, a Systematic Literature Review (SLR) was conducted using 115 published studies between 2010 and 2025 followed by empirical data collection from 30 Agile experts through semi-structured interviews conducted with practitioners from Agile companies and an online survey. This study applies Cumulative Voting (100-Dollar Test) and Multi-Criteria Decision Making (MCDM) techniques to rank and prioritize these challenges. Results: The SLR identifies several recurring Agile challenges; however, limited research has focused on their ranking and prioritization. The present study reveals new challenges, such as user interface complexities, lack of pre-development and pre-operational cost information, and lack of cost scalability at the module and feature levels. The current study identifies Inadequate Architecture (22%), Lack of Standardized Framework (18%), Communication and Coordination (16%), Poor Requirement Verification (13%), and Minimum Documentation (8%) as the most significant challenges. Conclusions: This study provides valuable insight for Agile practitioners and organizations, enabling more informed project planning, resource allocation, and strategic decision making. By focusing on the most critical challenges, teams can enhance software quality, streamline processes, and improve overall productivity in Agile environments. Full article

In the broader context of ecological transition, it is essential to identify solutions that ensure indoor environmental quality encompassing thermal, visual, acoustic, and indoor air quality conditions to safeguard occupant health and well-being. These solutions should also meet the demand for energy-efficient buildings. With specific regard to thermal environments, a distinction must be made between residential and non-residential settings, where comfort conditions can be achieved, and industrial environments, where thermal stress—and consequently health risks—may arise. To evaluate the quality of a thermal environment, key metrics are necessary. These include the Predicted Mean Vote (PMV) and the Predicted Percentage of Dissatisfied (PPD) for global thermal comfort, Predicted Heat Strain (PHS) and the Wet Bulb Globe Temperature (WBGT) for hot environments, and Required Insulation (IREQ) for cold environments, all governed by ISO-EN standards. The use of indices in residential and non-residential buildings outlines two critical challenges. The first relates to the fact that, in certain instances involving non-air-conditioned buildings, conditions can be borderline between comfort and thermal stress, which must be accurately identified. Secondly, the application of indices frequently neglects necessary variables, disregarding the fundamental limitations and operational boundaries inherent to both objective and personal input quantities. Moreover, the use of measurement devices inconsistent with the minimum requirements laid down by the standards in the field results in unwanted biases with unforeseeable consequences. This review explores the formulation, use, and limitations of the four indices mentioned, providing a perspective on their future development. It establishes the criteria for reliable long-term assessments of thermal and energy environments, encompassing the analysis of both heat and cold strain. Full article

Decarbonisation targets are now widely adopted. However, the speed at which renewables are built depends on many factors, including technology, cost, regulation, and the decisions of investors, developers, and governments. Public acceptance is one of these factors, but it is not the decisive one. This article examines public acceptance specifically: the attitudes, perceptions, and preferences of citizens and consumers, which form only one part of the wider social acceptance of renewable energy that also involves policy-makers, intermediaries, and market actors. We ask how public acceptance of renewable energy varies across countries, what drives that variation, and which policy instruments most reliably turn public approval into built capacity. Using a structured narrative synthesis across 21 countries, we integrate major cross-national opinion surveys (Eurobarometer 555, the UNDP–Oxford Peoples’ Climate Vote 2024, Pew, DESNZ, CSIRO, KfW, Ipsos), peer-reviewed work on local community acceptance, and capacity data from IRENA and the IEA. Three findings stand out. Headline support is high almost everywhere—a median of 76% across 77 countries and 85% in the EU—but increasingly masks polarisation, most sharply in the United States, where the partisan gap on prioritising renewables widened from 26 to 49 points between 2020 and 2024. Technology-specific acceptance follows a broadly consistent order, from rooftop solar down through wind, geothermal, hydropower, and biomass to carbon capture and nuclear. And community acceptance turns less on physical proximity to infrastructure than on procedural fairness, place attachment, and meaningful financial participation. Our contribution is integrative: we propose operational indicators of public acceptance that travel across national contexts, a country typology linking acceptance institutions to renewable outcomes, and a five-element policy architecture—early participation, benefit sharing, procedural transparency, place-sensitive siting, and credible information—each mapped to the acceptance dimension it addresses. Full article

Background/Objectives: Pain intensity and depressive symptoms are closely associated in adults with chronic pain (CP), contributing to greater disability, poorer treatment outcomes, and increased healthcare costs. However, the mechanisms underlying this symptom-level relationship remain incompletely understood. This systematic review and meta-analysis aimed to (i) identify all candidate factors mediating the effect of pain intensity on depressive symptoms, or vice versa, in adults with CP and (ii) estimate the magnitude of indirect, direct and total effects reported across eligible studies. Methods: Seven databases were searched in May–July 2023 and updated in May–June 2024. Observational studies and randomized controlled trials evaluating mediators of the relationship between pain intensity and depressive symptoms, or vice versa, in adults with CP were eligible. Data extraction and methodological appraisal were performed independently by three reviewers. Findings were synthesized narratively using vote counting and displayed using harvest plots. When appropriate, meta-analyses were conducted using regression coefficients, standard errors and sample sizes to estimate pooled indirect, direct and total effects. Results: The search identified 6826 studies, of which 34 (47 mediation models) met the inclusion criteria. Twenty-nine studies (combined n = 13,587) examined the effect of pain intensity on depressive symptoms, identifying 24 candidate mediators. Exploratory meta-analyses were feasible only for pain catastrophizing (indirect effect [IE]: β = 0.150; 95% confidence interval [CI]: 0.066, 0.233) and helplessness (IE: β = 0.101; CI: 0.066, 0.137). The pooled estimate for pain catastrophizing showed substantial heterogeneity and was highly influenced by a single study. Sleep quality, pain self-efficacy and pain interference showed preliminary evidence of mediation in narrative synthesis. Nine studies (combined n = 1359) investigated the effect of depressive symptoms on pain intensity, identifying 12 candidate mediators. Pain catastrophizing was the most consistently supported mediator, although quantitative synthesis was not feasible. Conclusions: Several potentially modifiable mediators of the pain–depression relationship were identified and may represent targets for multidisciplinary interventions. However, quantitative findings remain preliminary because of the limited number of studies, substantial heterogeneity for some mediators, and the predominance of cross-sectional evidence. Robust three-wave longitudinal mediation studies with adequate statistical power, standardized measures and reporting, and theory-driven models are needed to establish the temporal validity of candidate mediators and strengthen causal inferences. Full article

Hemophilia is an X-linked inherited bleeding disorder, classified as type A or type B. Therapeutic advances offer new treatment options that improve disease control and reduce associated complications, including inhibitor development and hemophilic arthropathy. This document aims to update the Mexican hemophilia consensus, reviewing current evidence on diagnosis and management, and addressing gaps in the treatment and follow-up of patients in Mexico, aligning local needs with international recommendations. A PubMed literature search covering the last five years (up to September 2025) was conducted, prioritizing consensus statements, guidelines, and systematic reviews. Using the Delphi methodology, a structured questionnaire was submitted electronically to forty-four experts. Aspects without initial agreement were discussed at an in-person meeting. Consensus was defined as at least 80% of votes in favor. Recommendations were issued across six domains: laboratory diagnosis, genetic testing, management of hemophilia A and B without and with inhibitors, adjuvant treatments, and gene therapy. The recommendations address prophylaxis with coagulation factor concentrates, non-factor therapies, immune tolerance induction, perioperative management, pain management, and eligibility criteria and follow-up protocols for gene therapy with adeno-associated viral vectors. This consensus provides updated, evidence-based recommendations adapted to the Mexican healthcare context, identifying priority areas, including timely access to non-factor therapies and gene therapy, development of a national referral network for complex cases, and inclusion of novel therapeutic agents in the institutional essential medicines list, with the aim of improving the quality of life of people with hemophilia in Mexico. Full article

Medical artificial intelligence systems often rely on a single model, which may give inconsistent answers and does not reflect the team-based nature of clinical consultation. This study develops a collaborative framework in which several agents analyze a medical question from different professional perspectives, discuss the evidence, vote on the proposed answer, and revise unresolved questions up to three times. Each step and stopping condition is recorded in a time-ordered workflow so that the decision process can be reviewed. The framework was evaluated on three regional versions of a medical examination dataset and three medical image question-answering datasets. It achieved an accuracy of 80.1% on the Mainland examination dataset and produced the highest reported results among the evaluated systems on ADAM. Stronger image models performed better on ACRIMA and Covid CT, showing that collaboration remains limited by the underlying visual model. Additional model-based quality scores are reported only as exploratory results because they were not validated by an independent model or human experts. These findings show that structured collaboration can make medical artificial intelligence workflows more transparent. Full article

Blockchain is a fast-changing field that is highly useful in such areas as finance, supply chain management, voting systems, and healthcare. As a consequence, software developers are increasingly creating Blockchain-Based Applications (BBAs) and Smart Contracts (SCs). However, the development of BBAs has been associated with various problems, especially in the process of updating and debugging such systems with a high degree of reliability. This is due to the immutability of deployed SCs. In this paper, we conduct an in-depth analysis of 61 published BBA articles between 2017 and 2025 to identify some causes of these challenges. Our results indicate that there is inadequate adaptation of the Software Development Life Cycle (SDLC) for BBAs. In particular, few BBA projects—only 32% of the reviewed projects—address the analysis phase, and only 29% deal with the design phase, frequently ignoring formal modeling methods. Based on these observations, we propose a new, context-adaptive methodology that facilitates BBA developers passing through the requirements, analysis, design, and implementation processes. Formal modeling techniques—such as Use Case Maps (UCMs), Finite State Machines (FSMs), and extended Unified Modeling Language (UML) class and sequence diagrams—are used within the methodology to document BBA structural and behavioral features and maintain complete traceability between requirements and implementation. In order to overcome the blockchain-specific drawbacks of traditional UML, we present formal stereotype extensions of UML class diagrams, where a four-compartment structure is introduced to differentiate state variables, functions, events, and access modifiers on SCs. We also provide analogous extensions to UML sequence diagrams using differentiated arrow notations to distinguish between function calls and event emissions to support accurate modeling of decentralized transaction flows. These extensions are described with a rationale and are formally defined and justified by mapping rules. Our methodology is justified by two case studies that prove its applicability in different fields of blockchain. The initial case study thus designs and executes a system of a halal chicken meat supply chain on Ethereum, showing the complete traceability of requirements that are based on UCM-based requirements and FSM-generated algorithms to implement SCs. The second case study applies the methodology to a decentralized Electronic Health Record (EHR) management system, and it shows coverage and completeness modeling. The methodology was evaluated through two case studies using a structured questionnaire and quantitative metrics, including traceability accuracy, reduction-in-error indicators, SC defect and gas-analysis results, modeling overhead measurements, and static security analysis with Slither. It is also evaluated based on a group of seven literature-based qualitative evaluation criteria that include workflow expressiveness, reusability, technical concept coverage, intelligibility, completeness, tool support, and blockchain limitation modeling. Full article

Background and Objectives: Persistent human papillomavirus (HPV) infection is associated with the development of cervical intraepithelial neoplasia and cervical cancer. In women with low-grade cervical lesions, a conservative management approach is commonly recommended. However, a subset of patients may experience lesion progression, which can also be accompanied by psychological distress. Adjuvant therapies during the surveillance period have gained increasing clinical interest. This consensus aims to provide multidisciplinary, expert-based recommendations on the use of a Coriolus versicolor-based vaginal gel in HPV-positive women with low-grade cervical lesions. Materials and Methods: A rapid systematic review was conducted in accordance with PRISMA guidelines, and the level of evidence was assessed using SIGN criteria. Subsequently, a multidisciplinary panel of experts formulated and evaluated consensus statements and recommendations using the nominal group technique and a voting process. Results: All statements reached consensus, with agreement levels exceeding 80%. Experts concluded that available evidence reports potential benefits of a Coriolus versicolor-based vaginal gel in HPV-positive women with low-grade cervical lesions, since it has demonstrated increased lesion regression and viral clearance rates. These benefits may be more pronounced in specific subgroups, particularly women over 40 years of age. Limited studies also suggest a positive effect on vaginal microbiota, which may contribute to both vaginal health and HPV clearance, as well as a reduction in perceived stress. The experts emphasized the importance of patient education, shared decision-making, and addressing psychosocial aspects, including emotional well-being and sexual health. Conclusions: This consensus provides a multidisciplinary perspective on the management of HPV-positive women with low-grade cervical lesions, highlighting the adjuvant role of a Coriolus versicolor-based vaginal gel during surveillance. Although current evidence suggests potential clinical benefits, further high-quality, independent studies are needed to corroborate these findings. Full article

Online business reviews encode rich customer sentiment that is critical for commercial decision making, yet accurately predicting star ratings from free text remains a challenging five-class classification problem. Classical ensemble methods—Soft Voting, Weighted Voting, and Stacking—aggregate complementary base-model outputs to improve predictive performance, but they produce opaque decisions that are unintelligible to business stakeholders. This paper proposes using a large language model (LLM), specifically unsloth/LLaMA-3.3-70B-Instruct, as an Explainable AI (XAI) ensemble aggregator: the LLM receives the predictions and confidence scores of four heterogeneous base models (Logistic Regression, Support Vector Machine, Naïve Bayes, and BERT-base-uncased) and reasons over them to produce both a final star-rating prediction and a natural-language explanation. We evaluate the full pipeline on 10,000-sample balanced and natural-distribution test sets derived from the Yelp Academic Dataset, with additional cross-lingual validation on Spanish Amazon Reviews. The LLM aggregator (LLAMA_AGG) achieves the highest macro-F1 on both pipelines (0.6800 on balanced; 0.6720 on natural) and the best ordinal calibration (QWK = 0.9111 on balanced; 0.9337 on natural), outperforming all classical aggregators and base models. A detailed Explainable AI analysis reveals that the LLM revises 28.07% of its standalone predictions after observing the ensemble outputs, improving the accuracy by +22.2 percentage points on the revised cases. The aggregator corrects severe polar bias in the standalone LLM (±0.35 recall improvement on mid-range star classes) and produces longer explanations when evidence is conflicted—a quantitative signal of deliberative reasoning. A formal human evaluation with two judges confirms high explanation faithfulness (4.47/5) and readability (4.82/5). Model scale ablation shows an 8B parameter variant achieves 90.8% agreement with the 70B model, enabling practical deployment. These findings demonstrate that Explainable AI can be achieved through LLM-based ensemble aggregation, establishing a principled approach for business-review sentiment analysis. Full article

Renewable energy communities (RECs) are legally recognised instruments of decentralised energy transition, but their scaling remains limited by a separation between governance design, economic appraisal, and multi-criteria decision analysis (MCDM/A). This systematic review applies a conservative explicit-reporting coupling diagnostic to 322 REC studies from a 2014–2026 corpus. The diagnostic identifies 267 unilateral studies (82.9%), 55 bilateral studies (17.1%), and no study satisfying the strict trilateral criterion at the title/abstract/metadata reporting threshold. A loosened-keyword sensitivity test identified two weak trilateral candidates, but the manual construct-validity audit did not support reclassifying them as strong trilateral studies. The result is therefore interpreted as a conditional and conservative signal of limited visible integration, not as proof that no body-text-level integration exists. The paper argues that the observed separation is not only methodological but also institutional. It therefore proposes second-level renewable energy cooperatives as federated decision-support entities that pool data stewardship, distributional economic appraisal, stakeholder-weighted MCDM/A, and auditable cooperative voting packages while preserving local REC control. The model converts the explicit-reporting gap into a testable governance architecture for REC scaling. Full article

Background/Objectives: The dental profession is undergoing significant transformation driven by societal changes, technological advancements, evolving patient expectations, and increased attention to sustainability. These developments challenge traditional notions of dental competence and highlight the need for educational frameworks that support adaptability and longitudinal professional development. The aim of this study was to develop a contextually grounded competence framework for undergraduate dental education through an iterative, multi-method process informed by key educational stakeholders. Methods: A multi-method approach was used, combining a preparatory phase (literature review, interviews) with a development phase (drafting and workshops) that was revisited in response to feedback, followed by iterative voting rounds that prompted further minor revisions. A deductive exploratory mapping analysis aligned the emerging framework with existing intended learning outcomes across the curriculum. Results: The multi-method process produced descriptions of a framework that deliberately integrates roles, skills, and attributes to capture key dimensions of professional competence in dentistry. The framework includes six domains: ‘Evidence-informed’, ‘Decision-maker’, ‘Communicator’, ‘Acting with professional conduct’, ‘Health ambassador’, and ‘Collaborator and leader’. Across voting rounds, the domains were generally rated between ‘neutral’ and ‘very important’, with each round prompting minor revisions. Mapping suggested alignment between the overarching framework and the detailed curriculum. Conclusions: This study presents the outcome of a structured, exploratory multi-method process to develop a locally relevant competence framework, integrated into a dental education. The participatory design supported clarity and relevance. While sharing similarities with existing frameworks, the new framework also includes differences. The term ‘professional conduct’ was preferred rather than ‘professionalism’, and the domains ‘collaborator and leader’ and ‘decision-maker’ were identified as relevant according to employer expectations. Although the work was based locally at one dental school, the approach may be transferable to similar contexts. Full article

Background: Pre-existing diabetes mellitus is prevalent among critically ill adults and can influence initial glycemic targets, therapeutic decisions, and early risk stratification in the intensive care unit (ICU). However, diabetes status may be distributed across heterogeneous electronic health record (EHR) sources and may be incomplete at the time of ICU admission, particularly for inter-facility transfers. Methods: Using the public WiDS Datathon 2021 tabular release derived from the Global Open-Source Severity of Illness Score (GOSSIS) initiative, we conducted a retrospective machine-learning benchmarking study for admission-time identification of documented diabetes status in ICU patients. Candidate predictors included demographics, admission characteristics, anthropometrics, day-1 physiologic and laboratory summaries, APACHE-related variables, comorbidity indicators, and site descriptors. We compared CatBoost, random forest, tuned XGBoost, tuned LightGBM, histogram-based gradient boosting, and a soft-voting ensemble combining XGBoost, LightGBM, and histogram-based gradient boosting. Because class imbalance was a central concern, the final workflow emphasized model-intrinsic class weighting and threshold-aware evaluation rather than synthetic oversampling. Results: In the primary leakage-mitigated random validation split, the voting ensemble achieved the highest overall balance, with AUROC 0.8539, precision 0.5671, recall 0.6690, and F1-score 0.6138. Tuned LightGBM was the most sensitivity-oriented individual model, achieving recall 0.7677 and AUROC 0.8537, although with lower precision and a less favorable Brier score. Ablation analyses clarified the source of this performance: removing leakage-prone and APACHE-related variables caused only modest decreases in discrimination, whereas the strict reduced model that also excluded glucose-like predictors produced a marked decline, with LightGBM AUROC falling to 0.7432 and the voting ensemble AUROC falling to 0.7448. These findings, together with SHAP analyses identifying day-1 glucose maximum, day-1 glucose minimum, BMI, age, hemoglobin, and related clinical variables as major contributors, indicate that glucose-related admission variables remained the dominant predictive signal. In grouped hospital validation, tuned LightGBM maintained recall of 0.7684 while AUROC decreased modestly to 0.8443, indicating preserved case detection under stricter site separation but reduced precision. Precision–recall analysis further showed that average precision decreased from 0.622 under random validation to 0.551 under grouped validation; at a high-sensitivity grouped-site operating point, a probability threshold of 0.4537 achieved recall of 0.8001 with precision of 0.4314. Calibration curves and Brier scores showed that predicted probabilities were imperfectly calibrated. Conclusions: Although the dominance of glucose-related predictors is clinically plausible for identifying documented diabetes status, early glycemic measurements in critically ill patients may also partly capture acute stress physiology, treatment-related effects, monitoring intensity, or other forms of acute dysglycemia rather than chronic diabetes status alone. Therefore, these findings support gradient-boosted and ensemble models as reproducible tools for ICU admission-time phenotyping of documented diabetes status, but the proposed system should be interpreted primarily as a screening-oriented phenotyping aid for chart review, cohort enrichment, or workflow support, not as a stand-alone diagnostic tool. Further external validation, recalibration, threshold selection matched to intended use, and clinical review are needed before deployment. Full article

Background/Objectives: Egypt ranks among the top ten countries globally with the highest burden of type 2 diabetes mellitus (T2DM), with prevalence projected to rise significantly by 2050. Despite multiple therapeutic options, glycemic control remains suboptimal due to therapeutic inertia, treatment complexity, and healthcare system limitations. Fixed-ratio combinations (FRCs) of basal insulin and glucagon-like peptide-1 receptor agonists (GLP-1 RAs) offer a simplified injectable strategy addressing complementary pathophysiological defects in T2DM. This study aims to develop expert consensus recommendations for the use of FRCs in Egyptian adults with T2DM, integrating international evidence with local practice. Methods: A modified Delphi technique was employed to achieve consensus among 9 diabetes experts across Egypt. Statements were formulated based on a targeted literature review and voted on using a structured Likert scale. Consensus was defined as ≥70% agreement. Results: Twenty-nine statements were endorsed with strong to very strong consensus. Recommendations covered patient selection, initiation after oral therapy or GLP-1 RA, switching from premixed or complex insulin regimens, dosing strategies, safety considerations, and intensification options. FRCs were favored for early injectable use, regimen simplification, and improved adherence, with liraglutide-based FRCs preferred for cardiovascular and renal benefits. Digital health integration was strongly recommended to enhance glycemic control and patient engagement. Conclusions: FRCs offer a simple and effective treatment simplification option for patients with uncontrolled T2DM on premix insulin, complex insulin regimens, or oral therapy. FRCs may improve glycemic control with generally favorable effects on hypoglycemia risk and body weight across many randomized and real-world studies, while reducing injection burden, simplifying the treatment regimen, and supporting patient adherence and satisfaction. Full article

Background/Objectives: Objective end-product assessment of microsurgical anastomoses is intensive and partly subjective. Vision-capable large language models (LLMs) may enable standardized image-based scoring, but their agreement with expert assessment remains uncertain. Methods: We studied 200 end-to-end femoral artery anastomoses, performed on chicken legs by novice, intermediate, and experienced microsurgeons. Images were scored independently by two blinded expert panels; disagreements were adjudicated by a third senior reviewer to establish expert consensus. Two LLMs, ChatGPT 5.2 Thinking Extended and Gemini 3.1 Pro, were evaluated using the exact same prompt and rubric. Each image was analyzed three times per model. Final scores were aggregated by median for numeric items and majority vote for categorical items. The primary endpoint was exact-match agreement with expert consensus. Agreement within ±1 was also assessed for numeric items. Agreement was measured using simple percentage agreement, Light’s kappa, and Krippendorff’s alpha; Bland–Altman analysis was used for numeric count items. Results: LLM 1 achieved a higher overall exact-match agreement than LLM 2 (0.659 vs. 0.539). Both models performed better on categorical than numeric items (0.713 vs. 0.610 and 0.651 vs. 0.445, respectively). LLM 1 showed the greatest advantages for gaps, knots, oblique stitches, and wide bites. Krippendorff’s alpha was positive for most endpoints with LLM 1, whereas LLM 2 showed negative values throughout. Allowing a ±1 tolerance for numeric items greatly improved agreement, suggesting only minor counting discrepancies, from 0.610 to 0.900 for LLM 1 and from 0.445 to 0.826 for LLM 2. Conclusions: Under a constrained scoring workflow, LLMs partially approximated intraluminal microsurgical end-product scoring. LLM 1 outperformed LLM 2, but agreement remained insufficient to replace the expert assessment entirely. These models can be assistive tools within a human-in-the-loop framework. Full article