Search Results (11,597)

This study aimed to determine the prevalence of the liver trematodes Fasciola spp. and Dicrocoelium dendriticum in cattle and sheep slaughtered in Siirt Province using fecal sedimentation and postmortem liver examination to compare their detection performance, and to estimate the sensitivity of fecal sedimentation relative to postmortem examination. A total of 2211 animals, comprising 374 cattle and 1837 sheep, slaughtered during the study period, were included in the study. Fecal samples were collected rectally before slaughter, and after slaughter, the liver, gallbladder, and bile ducts were examined macroscopically. Fecal samples were analyzed using sedimentation; postmortem examination was the reference method. Based on postmortem examination, Fasciola spp. and D. dendriticum were detected in 26.2% (98/374) and 7.2% (27/374) of cattle, and in 14.8% (271/1837) and 5.9% (108/1837) of sheep, respectively. Fecal examination detected significantly fewer infections for both parasites in both hosts (McNemar test, p < 0.001 for all comparisons). The sensitivity of fecal sedimentation was 80.6% for Fasciola spp. and 29.6% for D. dendriticum in cattle, and 86.7% and 73.1%, respectively, in sheep. The prevalence of Fasciola spp. was significantly higher in cattle than in sheep (OR = 2.05, 95% CI: 1.58–2.67; p < 0.001), whereas the difference in D. dendriticum prevalence between cattle and sheep was not statistically significant (OR = 1.24, 95% CI: 0.80–1.92; p = 0.274). In cattle, age older than two years was significantly associated with Fasciola spp. (adjusted OR = 15.82, 95% CI: 8.12–30.84; p < 0.001) and D. dendriticum infection (adjusted OR = 3.29, 95% CI: 1.36–7.96; p = 0.008). In sheep, the Morkaraman breed had lower odds of Fasciola spp. infection (adjusted OR = 0.42, 95% CI: 0.32–0.56; p < 0.001) and D. dendriticum infection (adjusted OR = 0.23, 95% CI: 0.14–0.38; p < 0.001) than the Akkaraman breed, while age was significantly associated only with D. dendriticum infection (adjusted OR = 1.84, 95% CI: 1.25–2.71; p = 0.002). These findings demonstrate that liver trematode infections were common among slaughtered animals in Siirt. Reliance on fecal sedimentation may considerably underestimate prevalence, particularly for D. dendriticum in cattle. Age- and breed-related associations should be considered in surveillance and control programs. From a public health perspective, awareness of zoonotic fasciolosis and risks associated with consuming raw aquatic plants from potentially contaminated areas should be increased. Full article

The spatial pattern of building damage produced by an earthquake carries information that classical building-by-building vulnerability indices cannot capture. This study presents one of the first frameworks to use Topological Data Analysis (TDA), a set of methods that quantify the “shape” of data, to characterise the spatial topology of seismic damage across an urban building inventory. Using the geo-referenced centroids of buildings as a point cloud, a sequence of connectivity graphs (a Vietoris–Rips filtration) is built at increasing distance scales, and persistent homology is used to track which spatial features appear and disappear. From this we extract four interpretable descriptors: Betti numbers (the numbers of connected building clusters and of enclosed gaps), persistence entropy (a measure of how disordered the damage pattern is), total persistence (the combined lifespan of all topological features), and the Wasserstein-2 distance (how far the post-earthquake pattern has moved from the intact pre-earthquake pattern). These descriptors form a physics-informed feature vector that is used to predict the building-cluster damage state. The developed framework was trained, tested, and validated on 1490 buildings over seven post-earthquake scenarios. Lognormal fragility parameters were estimated with maximum likelihood estimation, and an Artificial Neural Network (ANN) and a Random Forest (RF) were retrained on the same 593-building training dataset for comparison. On the 847-building benchmark, the TDA framework reached 93.3% accuracy (95% CI: 91.4–94.9%), F1 = 0.921 (0.902–0.940), and AUC = 0.933, using a stratified 70/15/15 split (training = 593, validation = 127, test = 127). This is a 6.0-percentage-point gain over the retrained ANN and a 12.1-percentage-point gain over the HAZUS-MH index (McNemar p = 0.017). Damage was recorded on the six EMS-98 states DS0–DS5, with DS4 and DS5 merged into a single class to give a five-class taxonomy, and building-type-specific inter-storey drift ratio thresholds were validated against EN 1998-3 (Eurocode 8 Part 3). Exact Rips computation is practical only for clusters up to about 2000 buildings; for larger populations, a CGAL (Computational Geometry Algorithms Library)-based sparse approximation with O(N log N) cost is recommended. It seems that the topological descriptions of the damage field may provide predictive information above and beyond that given by density and ground motion intensity and offer a reproducible tool for post-earthquake screening. Full article

Greenhouse gas emissions from heavy-duty vehicles (HDVs) must be drastically reduced. Battery electric trucks (BETs) are the main option for low-carbon road freight transport, but they require recharging infrastructure. However, a thorough cost analysis of public charging is lacking, especially for the Megawatt Charging System (MCS). This study estimates the infrastructure-related levelised cost of megawatt charging for battery electric trucks in Europe based on simulated truck operations and techno-economic modelling. The analysis combines empirical driving data with cost assumptions for MCS infrastructure. The reported values are infrastructure-only costs and include annualised capital expenditure, installation costs, grid connection costs and operating expenditure. They exclude electricity prices, taxes, levies, land costs and operator margins. Low- and high-cost scenarios differ in assumed charger hardware and installation costs, while grid connection costs and utilisation assumptions are held constant across scenarios. The results show that utilisation is the key driver of cost reductions over time. The infrastructure-related levelised cost of MCS declines to 0.03–0.07 EUR/kWh by 2050 under the analysed cost assumptions. The total annual infrastructure costs for Europe are estimated at 6.6–10.8 billion EUR, or 2.9–4.7 EUR cents/km. The results support policy decisions on infrastructure deployment and highlight the importance of coordinated rollout and demand growth. Full article

This study proposes a mathematical framework for predicting complete response-time distributions associated with parametric video-game configurations. Each configuration is encoded as a point in a normalized metric space, and the statistical descriptors of its response-time distribution (median, mean, selected quantiles, interquartile range, and Skewness) are treated as real-valued Lipschitz functions on that space. Predictions for unseen configurations are obtained through McShane–Whitney extension formulas, which provide geometrically controlled upper and lower bounds compatible with the empirical Lipschitz regularity of the observed data. To handle the sequential incorporation of new observations, a regularization mechanism is introduced that replaces raw descriptors violating Lipschitz continuity constraints with a convex combination of the observed value and a weighted historical estimate. In the extended version of the method, the coefficient of this combination is selected through a one-pass online Q-learning-inspired procedure that selects, for each descriptor and instability regime, a data-dependent trade-off between fidelity and geometric regularity. The final output is a continuous Log-Normal density fitted by nonlinear least squares to the predicted descriptors, together with a Wasserstein-type uncertainty band derived from the Lipschitz bounds. The framework is validated on a controlled experiment with 24 participants across 20 game levels. Results show that the predicted distributions shift systematically with the input configuration and that, in the illustrative comparison, the adaptive mechanism produces feature-specific smoothing decisions that differ from those obtained with a fixed coefficient. The method also provides interpretable predictions from small experimental datasets without requiring fully data-driven models. Full article

Background: High tibial osteotomy (HTO) is commonly performed to manage medial compartment knee osteoarthritis by correcting mechanical alignment; however, the role of adjunctive regenerative therapies remains uncertain. Methods: This retrospective study compared leukocyte-rich platelet-rich plasma (LR-PRP) with leukocyte-poor PRP (LP-PRP) in patients undergoing HTO. Forty patients were allocated into three groups: HTO alone (n = 10), HTO with LR-PRP (n = 20), and HTO with LP-PRP (n = 10). Clinical outcomes were assessed preoperatively and at 12 months using the Visual Analog Scale, Oxford Knee Score, and Western Ontario and McMaster Universities Osteoarthritis Index. Cartilage repair appearance was evaluated through second-look arthroscopy using the ICRS grading and Koshino staging systems. Multivariable analysis of covariance (ANCOVA), adjusting for baseline imbalances, was employed to evaluate postoperative outcomes. Results: All groups demonstrated significant improvements in pain and function (p < 0.05), with no significant differences among groups. However, Group B exhibited a greater shift toward lower ICRS grades than Group A (p < 0.05), whereas no significant difference was found between Groups C and A. Arthroscopic findings revealed more complete defect coverage and improved structural integrity in the LR-PRP group. Conclusions: These findings demonstrate a clear discrepancy exists between clinical and structural outcomes; while HTO drives substantial and comparable short-term functional improvements across all cohorts, adjunctive LR-PRP is positively associated with a significantly enhanced arthroscopic cartilage repair appearance compared to LP-PRP or HTO alone. Further prospective studies are needed to validate these findings and elucidate the underlying biological mechanisms. Full article

Task decomposition is a fundamental challenge in multi-agent collaborative maintenance systems, where unstructured natural language instructions must be precisely translated into logically coherent, executable sub-task sequences. This paper formulates task decomposition as a constrained optimal path search problem on a heterogeneous knowledge graph that encodes coal mine equipment topology, fault causality, and maintenance procedures. We construct a composite cost function that systematically integrates semantic similarity from graph neural network embeddings, relation-type weights, and structural path length, transforming instruction parsing into a mathematically tractable combinatorial optimization. The cost function is derived from the principles of shortest-path reasoning in knowledge graphs: the relational weights capture domain-specific association strengths, the semantic similarity term promotes contextually coherent chains, and the path-length penalty prevents unnecessarily long derivations. A multi-hop reasoning algorithm coupling heterogeneous graph convolution with beam search is developed to solve this problem efficiently, achieving high-quality approximate solutions while ensuring computational tractability. The reasoning process is inherently interpretable, as the optimal path directly maps to a traceable atomic task sequence with explicit dependency relations. A formal complexity analysis shows the algorithm scales as $O$($b$·$K$·$d_{max}$), where b is the beam width and d_max is the maximum node degree. Several theoretical properties of the proposed framework are further derived: the cost function is non-negative and strictly monotonic, optimal paths satisfy the optimal substructure property, cycle-free optimal paths always exist, and beam search can yield globally optimal solutions given a sufficiently large beam width. These theoretical conclusions establish mathematical guarantees for the presented decomposition framework. Experiments on 200 composite maintenance instructions with gold-standard annotations (inter-annotator agreement Cohen’s $\kappa =0.88$) demonstrate that the proposed method achieves 94.3% task sequence accuracy (95% CI: 91.2–96.8%) and 96.4% dependency accuracy (95% CI: 93.5–98.1%), substantially outperforming both a rule-augmented baseline (58.6%, 62.1%) and a GPT-4o few-shot chain-of-thought baseline (73.2%, 70.5%); McNemar’s test yields p < 0.001 for both comparisons. The average inference time is 29.7 ms (SD 2.1 ms), meeting stringent industrial real-time constraints. Ablation studies quantify the contribution of each cost function component and confirm the robustness of the chosen beam width and hyperparameters. When integrated into a full multi-agent system, the framework delivers end-to-end response time within 3 s (P50: 1.87 s, P95: 2.83 s) and maintains an 86.7% task success rate even under dual agent failures, validating the robustness of the proposed mathematical formulation. This work establishes a rigorous graph-theoretic foundation for instruction decomposition in multi-agent systems, with direct applicability to safety-critical industrial environments. Full article

Background: Post-acute sequelae of COVID-19 (PASC), commonly known as long COVID, affects an estimated 10–30% of SARS-CoV-2 non-hospitalised and 50–70% of hospitalised survivors. This condition remains clinically heterogeneous, and the specific mechanisms driving the transition from acute infection to chronic sequelae remain poorly understood. We assessed independent risk factors, tracked the evolution of clinical features, defined distinct symptom-based phenotypes, and assessed the impact of different pandemic waves on the likelihood of developing long COVID in hospitalised survivors. Methods: We conducted a single-centre, retrospective cohort study at a university hospital in London. The population comprised 627 adults hospitalised with acute COVID-19 between February 2020 and December 2022. Baseline characteristics and outcomes were compared between long COVID and resolved cases using appropriate statistical tests for continuous and categorical variables. Multivariable logistic regression identified risk factors. McNemar’s test quantified the phenotypic shift from admission to follow-up. Latent Class Analysis (LCA) identified clinical subtypes based on symptom clusters. Results: Of 627 patients, 252 (40.2%) met long COVID criteria. Comorbidity burden was the strongest predictor; patients with a single condition had a 4-fold increase in odds (aOR 4.62, 95% CI 2.36–9.04). The ORs were also significantly elevated among patients with 2 (aOR 3.26), 3 (aOR 2.68), or 4 or more (aOR 3.24) comorbidities. Older age (aOR 1.04), acute disease severity (measured by length of hospital stay) (aOR 1.27 per log-day) and elevated admission fibrinogen (aOR 1.21 per g/L) were significant predictors. Temporal analysis revealed a precipitous decline in risk from Wild-type/Alpha (>50%) to Delta/Omicron (<21%). We observed a distinct phenotypic shift: while acute respiratory inflammation resolved, systemic fatigue increased fourfold (7.0% to 32.4%), and memory difficulties emerged in the post-acute phase. LCA identified two phenotypes: fatigue-dominant and multisystem phenotypes. Conclusions: Long COVID is a multifactorial syndrome driven by host susceptibility, acute severity, and persistent coagulopathy. Clinical management should move beyond a monolithic approach and favour phenotype-specific strategies. Full article

Background: Functional somatic disorder (FSD) is prevalent, but pathological pathways have not yet been identified. Cross-sectional studies suggest an association between FSD and lipid and glucose metabolism. The aim of the present prospective study was to assess whether metabolic factors represent possible risk factors for development of FSD. Methods: A population-based cohort of 9656 men and women aged 18–76 years was established in 2011–2015, and 5738 participated in a re-examination after a median of 65 months. At both baseline and follow-up, participants answered questionnaires on lifestyle and various delimitations of FSD including bodily distress syndrome (BDS), chronic fatigue (CF), chronic widespread pain (CWP), irritable bowel (IB), and multiple chemical sensitivity (MCS). Fasting values of cholesterol, triglycerides, glucose, HbA1c, insulin, and insulin resistance were assessed at baseline and analyzed in relation to incidence of FSD using logistic regression models, adjusted for age, sex, lipid- and glucose-lowering medications, obesity, social status, lifestyle, and sleep quality. Results: Several small, although significant, effects were observed. Lower levels of cholesterol were associated with development of BDS (OR: 0.982 (95% CI: 0.971–0.993)), CF (OR: 0.980 (95% CI: 0.964–0.996)), and CWP without co-morbid FSS among men. Higher levels of triglycerides were associated with development of CF (OR: 1.022 (95% CI: 1.003–1.041)). Glucose and HbA1c levels were not associated with development of FSD. Insulin and HOMA-IR levels were associated with development of CF in a non-linear relation, being negative at lower levels and changing to a positive relation at higher levels. Conclusions: More large-scale cohort studies are necessary to explore whether metabolism could play a role in the pathological pathway of FSD as indicated in the present study. Full article

Background: Cervical cancer screening uptake in Romania remains below 20%, with invasive sampling cited as a major participation barrier. Urinary HPV testing is a non-invasive alternative, but data from Eastern European populations are scarce and reporting stratified by histological grade is inconsistent. Objectives: To evaluate the agreement between paired urinary and clinician-collected cervico-vaginal HPV testing, and to assess the diagnostic performance of urinary HPV testing for biopsy-confirmed cervical intraepithelial neoplasia grade 2 or worse (CIN2+). Methods: Prospective cross-sectional study in three outpatient obstetrics and gynecology clinics (May 2025–May 2026). A total of 230 women aged 25–70 years provided paired cervico-vaginal (PreservCyt^®) and first-void urine (Colli-Pee^®) specimens. To capture the full spectrum of disease probability, participants were recruited across three clinical scenarios: primary screening, cytology triage and HPV-positive triage. A multiplex RT-PCR assay targeted a total of 14 high-risk HPV genotypes, providing separate identification for HPV16 and HPV18, alongside a pooled detection for the remaining 12 high-risk strains. Colposcopy-guided cervical biopsy served as the reference standard. Results: Paired hrHPV testing achieved substantial agreement (κ = 0.696, 95% CI 0.604–0.788), with higher genotype-specific concordance for HPV16 (κ = 0.755) and HPV18 (κ = 0.789). Discordance was directionally asymmetric (30 cervical-positive/urine-negative versus 4 urine-positive/cervical-negative; McNemar p < 0.001). For biopsy-confirmed CIN2+ (59 of 230; 25.7%), urinary hrHPV showed a sensitivity of 86.4% (95% CI 75.5–93.0), specificity 56.7%, positive predictive value 40.8% and negative predictive value 92.4%, compared with 91.5%, 43.3%, 35.8% and 93.7% for cervico-vaginal testing. Urinary HPV positivity increased monotonically with histological severity (25.0% no-lesion, 69.0% CIN1, 86.4% CIN2+; linear-by-linear p < 0.001). Conclusions: Urinary HPV testing demonstrates substantial agreement with clinician-collected sampling, near-equivalent negative predictive value, and a robust dose–response with histological severity. It is a clinically credible non-invasive entry point for a triage cascade in settings such as Romania, where low participation rather than analytical performance is the principal screening barrier. Full article

This work analyzes the ergodic capacity behavior of reconfigurable intelligent surface (RIS)-assisted multiple-input multiple-output (MIMO) systems with a finite and arbitrary number of antennas and RIS elements under Nakagami-m fading conditions. By combining Hadamard’s determinant inequality with the Cauchy–Schwarz inequality, this work derives a dimensionally consistent closed-form upper bound on the ergodic capacity in terms of the Meijer G-function. Subsequently, it is demonstrated that at a high signal-to-noise ratio (SNR), a simplified expression for the capacity upper bound can be derived, enabling an analytical assessment of how the fading parameter influences the ergodic capacity. The study also explores the asymptotic behavior in the large-system regime, where the number of antennas or RIS elements tends to infinity. Monte Carlo (MC) simulations confirm the accuracy of the proposed bound and scaling laws. Full article

Background/Objectives: Multiple validated microsurgical assessment instruments exist that differ in design and theoretical approach. Product-based tools evaluate the quality of the completed anastomosis, while process-based tools assess technique during task execution. Whether these tools measure the same underlying construct and can be used interchangeably has not been directly examined. This study compared the behaviour of three validated instruments, ALI, MARS10, and SMaRT, when applied to identical performances. Methods: Forty-five participants from 26 institutions across 14 countries performed an eight-stitch end-to-end vascular anastomosis on a chicken leg model, independently evaluated by two trained graders using all three instruments. Interrater reliability was assessed using intraclass correlation coefficients (ICC(2,1)), rank-order agreement using Spearman’s rho and Kendall’s W, and score distributions were examined for floor and ceiling effects. Results: All three instruments demonstrated high interrater reliability (ICC: ALI 0.955, MARS10 0.939, SMaRT 0.960), though product-based tools showed wider confidence intervals. Systematic rater differences were observed across all instruments. Rank correlations were strongest between ALI and MARS10 ($\rho$ = 0.913), with moderate correlations between process-based and product-based tools ($\rho$ = 0.698–0.749). MARS10 showed a ceiling effect, SMaRT underutilised its upper range, and ALI had the greatest score variability. Conclusions: Although demonstrating high interrater reliability, the instruments differ in rater variability, discriminatory capacity, and the aspects of performance prioritised, embodying different definitions of surgical skill. Current tools are not interchangeable, and instrument selection carries potential consequences for training and research. Future research in technology-driven approaches could offer a path toward more objective and consistent evaluation that is less dependent on human interpretation. Full article

High-density, intensive aquaculture has led to the deterioration of the aquatic environment, which compromises the survival, immune function and disease resistance of aquatic animals. This study investigated the effects of dietary supplementation with different levels of Macleaya cordata extract (MCE) on growth performance, muscle composition, antioxidant capacity, immune function, and stress resistance in red claw crayfish (Cherax quadricarinatus). Juvenile crayfish with an initial body weight of 0.22 ± 0.02 g were fed diets supplemented with 0 (Diet 1, Control), 0.01 (Diet 2), 0.02 (Diet 3), 0.04 (Diet 4), 0.08 (Diet 5), and 0.15 (Diet 6) g/kg MCE for eight weeks. Results showed that dietary MCE supplementation had no significant effect on the growth performance or muscle composition of crayfish. Conversely, supplementation with 0.02–0.04 g/kg MCE significantly increased the activities of superoxide dismutase (SOD), glutathione peroxidase (GPx), phenoloxidase (PO), and the total antioxidant capacity (T-AOC) in the hepatopancreas and plasma, while significantly reducing malondialdehyde (MDA) levels. Specifically, supplementation with 0.04 g/kg MCE significantly upregulated the hepatopancreatic mRNA expression levels of SOD, hemocyanin, GPx, and selenium-dependent glutathione peroxidase (Se-GPx). Furthermore, the challenge test demonstrated that supplementation with 0.04 g/kg MCE significantly improved crayfish survival under microcystin-LR (MC-LR) stress. Overall, supplementation with 0.04 g/kg MCE significantly improved the antioxidant capacity, immunity, and stress resistance of crayfish, suggesting its potential as an environmentally friendly immunostimulatory feed additive to protect against MC-LR stress in crayfish. Full article

Spectral analysis technology has emerged as a vital tool for quantifying complex samples because of its simplicity and high efficiency. However, spectral data has a high-dimensional characteristic and traditional variable selection methods struggle to balance computational efficiency and prediction accuracy. Hence, a discretized wild horse optimizer (WHO) algorithm was introduced in this study. Firstly, transfer functions were introduced to solve the discrete optimization in spectral variable selection. The optimal number of latent variables (LVs) in partial least squares (PLS), DWHO iterations and the population size were determined to establish the DWHO-PLS quantitative analysis model. Then, three spectral datasets of pork, marzipan and DOSY samples were used to assess the effectiveness of the method. Finally, the DWHO-PLS model was compared with full-spectrum PLS, uninformative variable elimination-PLS (UVE-PLS), Monte Carlo-UVE-PLS (MC-UVE-PLS), randomization test-PLS (RT-PLS), gray wolf optimizer-PLS (GWO-PLS) and whale optimization algorithm-PLS (WOA-PLS). Results show that the number of variables selected by DWHO-PLS was the smallest and the root mean squared error of prediction (RMSEP) had the lowest value compared with other methods for the three datasets. The research indicates that DWHO can effectively simplify the PLS model while enhancing its accuracy and stability. Full article

Generative Artificial Intelligence (GenAI) is rapidly reshaping higher education, especially in technology-oriented programs where critical thinking and complex problem solving are core outcomes. This study triangulates global and local evidence on performance/efficiency, usage, autonomy, critical-thinking engagement, and ethics by combining a systematized review informed by Kitchenham and structured using selected PRISMA 2020 elements (2020–2025; last search: May 2025; 49 studies; Scopus, ACM Digital Library, IEEE Xplore, and SpringerLink; not prospectively registered) with an anonymous survey of 302 computing and engineering students from a single university in Ecuador. The expert-reviewed instrument showed acceptable internal consistency for most scale-based dimensions (McDonald’s $\omega$), whereas institutional and ethics-related items were analyzed primarily at the item level. Results showed near-universal academic GenAI use (96%), with 47% of students reporting weekly use and 26% daily use. Research-related work was the most frequent application (81.5%), followed by homework (48.3%), report writing (43.7%), and exam preparation (41.7%). Although students reported perceived efficiency gains, concerns persisted about reduced analytical engagement and technological dependence (84.1%). Ethical concerns centered on dependence, authenticity, and data privacy, while institutional responses pointed to the need for formal training (96.7%) and clearer guidance. Based on this triangulation, we propose a context-bounded interpretive framework suggesting that GenAI’s educational value depends on instructional and governance conditions that preserve autonomy, critical thinking, integrity, and equity. Full article

Background: Korean-language medical question-answering datasets are increasingly used for large language model (LLM) development, but structural completeness alone does not establish model performance or clinical validity. We examined two national Korean medical knowledge datasets by combining a public labeling-file audit with a secondary evaluation of item-level outputs distributed with the corresponding official LLM packages. Methods: We reviewed the official documentation and complete public download inventories, parsed 34 training/validation labeling archives containing 31,036 records, and audited required fields, duplication, text length, question type, and specialty distribution. We also audited the official Qwen2.5-14B LoRA model packages and independently recalculated performance from their distributed item-level KorMedMCQA output files (2494 identical items per model). Accuracy was reported with Wilson 95% confidence intervals; model outputs were compared using an exact McNemar test and a paired bootstrap confidence interval. Because no compatible local GPU was available, model inference was not independently rerun. Results: The documentation described 34,487 labeled QA pairs, of which 31,036 (90.0%) were present in the publicly accessible training/validation labeling files; no public test-labeling archive was listed. Required fields were complete, no duplicated qa_id values were found, and one duplicated question-answer pair occurred in the Essential dataset. Multiple-choice items comprised 78.7% of all documented QA pairs, and the top three domains comprised 56.8%. In the distributed KorMedMCQA outputs, the Essential-care model answered 1603/2494 items correctly (64.27%; Wilson 95% CI 62.37–66.13%), while the Specialized-medicine model answered 1596/2494 correctly (63.99%; 95% CI 62.09–65.85%). The paired difference was 0.28 percentage points (bootstrap 95% CI −0.44 to 1.00), with no significant difference by exact McNemar test (46 vs. 39 discordant correct items; p = 0.515). Conclusions: The public training/validation labeling files showed favorable basic structural completeness, but the unavailable test labels, multiple-choice predominance, domain imbalance, limited record-level provenance, and incomplete model-package traceability constrain claims of clinical readiness. The distributed model outputs demonstrated moderate examination-style benchmark performance without a significant difference between the two models. These findings support use as research infrastructure, not evidence of clinical validity, and indicate the need for independent inference reproduction, clinician-led open-ended and safety evaluation, temporal updating, specialty-stratified reporting, and human oversight before clinical use. Full article