Search Results (16,679)

Auditability and regulatory compliance are increasingly required for deploying large language models (LLMs). Prior work typically targets isolated stages such as training or unlearning and lacks a unified mechanism for verifiable accountability across model updates. This paper presents AuditableLLM, a lightweight framework that decouples update execution from an audit-and-verification layer and records each update as a hash-chain-backed, tamper-evident audit trail. The framework supports parameter-efficient fine-tuning such as Low-Rank Adaptation (LoRA) and Quantized LoRA (QLoRA), full-parameter optimization, continual learning, and data unlearning, enabling third-party verification without access to model internals or raw logs. Experiments on LLaMA-family models with LoRA adapters and the MovieLens dataset show negligible utility degradation (below 0.2% in accuracy and macro-F1) with modest overhead (3.4 ms/step; 5.7% slowdown) and sub-second audit validation in the evaluated setting. Under a simple loss-based membership inference attack on the forget set, the audit layer does not increase membership leakage relative to the underlying unlearning algorithm. Overall, the results indicate that hash-chain-backed audit logging can be integrated into practical LLM adaptation, update, and unlearning workflows with low overhead and verifiable integrity. Full article

Background: Semaglutide is a Glucagon-like peptide-1 receptor agonist (GLP-1RA) used in the treatment of type 2 diabetes mellitus (T2DM) and weight management. While its clinical benefits are well established, concerns have emerged over off-label use, underreporting of adverse drug reactions (ADRs), and prescribing disparities. Aims: To examine real-world prescribing pattern treatment efficacy and ADRs associated with semaglutide in a socioeconomically deprived United Kingdom (UK) locality, and to compare these with national data. Methods: A retrospective service evaluation was conducted using anonymised data from 1403 patients across 42 GP practices under a data share agreement across a place-based group of practices in the West Midlands. National prescribing data were obtained from OpenPrescribing, and ADR data from the Medicines and Healthcare products Regulatory Agency (MHRA) Yellow Card Scheme (01/2020–12/2024). Analyses included demographic trends, treatment efficacy (HbA1c and weight), and socioeconomic comparisons using the Socioeconomic Index for Small Areas (SEISA) deciles. Findings: Semaglutide prescribing in the GP surgeries studied peaked in 2022 and declined thereafter, mirroring national trends. Prescribing of semaglutide mirrored the ethnic make-up of the region studied with a notable exception of White British. Mean HbA1c fell by 10.8 mmol/mol and weight by 4.8%. ADR incidence in the population studied (1.85%) exceeded national reporting rates (0.20%). Prescribing was highest in practices serving the most deprived communities. Conclusions: Semaglutide is effective in reducing HbA1c and weight in real-world settings. However, ADRs remain underreported. Socioeconomic deprivation was strongly associated with higher prescribing rates. Greater attention to equitable access and pharmacovigilance is warranted. Full article

The development of highly sensitive approaches for detecting tumor cells in biological samples remains a critical challenge in laboratory and clinical oncology. In this study, we investigated the structural and magnetic properties of iron oxide nanoparticles incorporated into cellulose microspheres of two size ranges (~100 and ~700 μm) and evaluated their potential for targeted tumor cell isolation. In the smaller microspheres, magnetite-based magnetic nanoparticles (MNPs) were synthesized in situ via co-precipitation, whereas pre-synthesized MNPs were embedded into the larger microspheres. The geometrical characteristics of the resulting magnetic cellulose microspheres (MSCMNs) were assessed by confocal microscopy. Transmission electron microscopy and X-ray diffraction analyses revealed an average magnetic core size of approximately 17 nm. Magnetic properties of the MNPs within MSCMNs were characterized using a highly sensitive nonlinear magnetic response technique, and their dynamic parameters were derived using a formalism based on the stochastic Hilbert–Landau–Lifshitz equation. To evaluate their applicability in cancer diagnostics and treatment monitoring, the MSCMNs were functionalized with a TKD peptide that selectively binds membrane-associated Hsp70 (mHsp70), yielding TKD@MSCMNs. Magnetic separation enabled the isolation of tumor cells from biological fluids. The specificity of TKD-mediated binding was confirmed using Flamma648-labeled Hsp70 and compared with control alloferone-conjugated microspheres (All@MSCMNs). The ability of TKD@MSCMNs to selectively extract mHsp70-positive tumor cells was validated using C6 glioma cells and mHsp70-negative FetMSCs controls. Following co-incubation, the extraction efficiency for C6 cells was 28 ± 14%, significantly higher than that for FetMSC (7 ± 7%, p < 0.05). These findings highlight the potential of TKD-functionalized magnetic cellulose microspheres as a sensitive platform for tumor cell detection and isolation. Full article

Background: The management of acute pulmonary embolism (PE) in the Emergency Department (ED) remains challenging, particularly in hemodynamically and respiratory stable patients with minimal symptoms. Diagnostic and therapeutic difficulties are further compounded when the condition is complicated by a mobile right atrial (RA) thrombus, representing an extreme-risk phenotype. Case Presentation: We report the case of a 65-year-old male with a single known venous thromboembolism risk factor-chronic venous insufficiency-who presented to the ED following a transient episode of severe dyspnea at home. On admission, he was hemodynamically and respiratory stable, without the need for oxygen supplementation. Arterial blood gas analysis revealed a metabolically compensated acidosis with elevated lactate, while cardiac biomarkers were moderately increased. Emergency point-of-care transthoracic echocardiography (POCUS-TTE) demonstrated severe right ventricular (RV) dysfunction and a large, mobile intracardiac thrombus prolapsing through the tricuspid valve. Computed Tomography Pulmonary Angiography confirmed pulmonary embolism and revealed a massive and extensive bilateral thrombotic burden (Qanadli score 32 points). Given the extreme risk for fatal embolization, immediate full-dose systemic thrombolysis with Alteplase (100 mg over 2 h) was initiated in the ED. Thrombolysis was completed without hemorrhagic complications. Follow-up POCUS-TTE at 2 h showed complete resolution of the intracardiac thrombus and significant improvement of RV function (RV/RA gradient reduced from 40 mmHg to 28 mmHg). Conclusions: This case highlights the effectiveness and safety of early systemic thrombolysis guided by ED POCUS-TTE in PE with a massive thrombotic burden, complicated by a mobile intracardiac thrombus, even in the absence of shock. Such prompt intervention may reduce mortality risk in intermediate-to-high-risk PE subsets, despite limited guidance in current clinical recommendations. Full article

The Satellite Internet of Things (IoT) sector is undergoing rapid transformation, driven by breakthroughs in satellite communications and the pressing need for seamless global coverage—especially in remote and poorly connected regions. In locations where terrestrial infrastructure is limited or non-existent, Low Earth Orbit (LEO) satellites are proving to be a game-changing solution, delivering low-latency and high-throughput links well-suited for IoT deployments. While North America currently dominates the market in terms of revenue, the Asia-Pacific region is projected to lead in growth rate. Nevertheless, the development of satellite IoT networks still faces hurdles, including spectrum regulation and international policy alignment. In this evolving landscape, the LoRa and LoRaWAN protocols have been enhanced to support direct communication with LEO satellites, typically operating at altitudes between 500 km and 2000 km. This paper offers a comprehensive review of current research on LoRa/LoRaWAN technologies integrated with LEO satellite systems, also providing a performance assessment of this combined architecture in terms of theoretical achievable bitrate, Bit Error Rate (BER), and path loss. The results highlight the main performance trends of LoRa LR-FHSS in direct-to-LEO links. Path loss increases sharply with distance, reaching approximately 150 dB at 500 km and 165–170 dB at 2000 km, significantly reducing achievable data rates. At 500 km, bitrates range from approximately 7–8 kbps for SF7 to below 2 kbps for SF12. BER follows a similar trend: below 200 km, values remain low (${10}^{-4}$–${10}^{-3}$) for all spreading factors. At 1000 km, BER rises to approximately $3.9\times {10}^{-3}$ for SF7 and $1.5\times {10}^{-3}$ for SF12. At 2000 km, BER reaches approximately $4.7\times {10}^{-2}$ for SF7 but stays below $2\times {10}^{-2}$ for SF12, showing a 2–$3\times$ improvement with higher spreading factors. Overall, many links exhibit path loss above 160 dB and BER in the ${10}^{-3}$–${10}^{-2}$ range at long distances. These results underscore the importance of adaptive spreading factor selection and LR-FHSS gain for reliable long-range satellite IoT connectivity, highlighting the trade-off between robustness and spectral efficiency. Full article

Exploring green organic solvents is a global demand. Most of the currently used solvents pose some concerns regarding environmental sustainability and occupational health risks. In this work, propylene glycol was employed for the first time as a green solvent for mobile phase preparation in the reversed phase chromatographic separation of a mixture of two antioxidants, glutathione and ascorbic acid. The slight viscosity of propylene glycol was manipulated by using water as a co-fluidizing agent to facilitate pumping. Method optimization was performed using factorial design experimental Expert 13^® Software (Minneapolis, MN, USA) to achieve the maximum resolution and the minimum run time. The reported method was properly validated according to the International Conference on Harmonization criteria at the linearity range of 1–500 µg/mL, with acceptable accuracy and precision for both drugs. The method was effectively applied for the quantification of both drugs in their commercial pharmaceutical formulation. The proposed method was assessed for environmental and operator safety by means of global tools like AGREE and MoGAPI and has proved high degrees of greenness. Propylene glycol has several benign properties, such as low volatility, less toxicity, compatibility with UV detectors and very low flammability, that will soon assemble it as a promising alternative for the conventionally used solvents. Full article

GC-MS oxylipin profiling of brown and red algal thalli was performed. Brown algae (Fucus distichus and Alaria esculenta) were collected from the Barents Sea coastline nearby Teriberka, Murmansk region, Kola Peninsula, Russia, while other brown and red algae were sourced from the Pacific coast of the Russian Far East. Triols and δ-ketols (epoxyalcohol synthase products) were found in most brown and red algae. Several Heterokontophyta and Rhodophyta species possessed α-ketols (products of allene oxide synthase) and related vic-diols. Plasmodiophorols and ectocarpins (hydroperoxide bicyclase (HPB) products) were found only in brown algae from the Ectocarpales, Fucales, and Laminariales orders, not in brown algae from the Desmarestiales or Dictyotales orders, or in any red algae. Therefore, plasmodiophorol A and other HPB products could be used as chemotaxonomic markers for the classification of the separate orders of algae within Heterokontophyta. The in vitro incubations of F. distichus thalli with linoleic and α-linolenic acid resulted in the formation of α-ketols and the hydroperoxide bicyclase product, plasmodiophorol A. Full article

Understanding the spatial and temporal structure of crustal deformation is essential for identifying tectonic blocks, assessing seismic hazard, and detecting precursory deformation associated with major megathrust earthquakes. In this study, we analyze twenty years of continuous GNSS observations from the Japanese Islands to identify coherent deformation domains and anomalous regions using an integrated time-dependent clustering framework. The workflow combines six machine learning algorithms (Hierarchical Agglomerative Clustering, K-means, Gaussian Mixture Models, Spectral Clustering, HDBSCAN and consensus clustering) and constructs a set of deformation-related features including steady-state velocities, strain rates, co-seismic and post-seismic displacements, and spatial distance metrics. Optimal cluster numbers are determined by validity metrics, and the most robust segmentation is obtained using a consensus approach. The resulting spatiotemporal domains reveal clear segmentation associated with major geological structures such as the Fossa Magna graben, the Median Tectonic Line, and deformation belts related to Pacific Plate subduction. The method also highlights deformation patterns potentially associated with the preparation stages of megathrust earthquakes. Our results demonstrate that machine learning-based clustering of long-term GNSS time series provides a powerful data-driven tool for quantifying deformation heterogeneity and improving the understanding of active geodynamic processes in subduction zones. Full article

This study assesses the suitability of UAV aerial imagery-based photogrammetry for reconstructing underwater riverbed topography and its application in two-dimensional (2D) hydrodynamic modelling, with a particular focus on comparing RGB, multispectral, and fused RGB–multispectral imagery. Four Lithuanian rivers—Verknė, Šušvė, Jūra, and Mūša—were selected to represent a wide range of hydromorphological and hydraulic conditions, including variations in bed texture, vegetation cover, and channel complexity. High-resolution digital elevation models (DEMs) were generated from field-based surveys and UAV imagery processed using Structure-from-Motion photogrammetry. Two-dimensional hydrodynamic models were created and calibrated in HEC-RAS 6.5 using measurement-based DEMs and subsequently applied using photogrammetry-derived DEMs to isolate the influence of terrain input on model performance. The results showed that UAV-derived DEMs systematically overestimate riverbed elevation, particularly in deeper or vegetated sections, resulting in underestimated water depths. RGB imagery provided greater spatial detail but was more susceptible to local anomalies, whereas multispectral imagery produced smoother surfaces with a stronger positive elevation bias. The fusion of RGB and multispectral imagery consistently reduced spatial noise and improved hydrodynamic simulation performance across all river types. Despite moderate vertical deviations of 0.10–0.25 m, relative flow patterns and velocity distributions were reproduced with acceptable accuracy. The findings demonstrate that combined spectral UAV aerial imagery in photogrammetry is a robust and cost-effective alternative for hydrodynamic modelling in shallow lowland rivers, particularly where relative hydraulic characteristics are of primary interest. Full article

Reproductive traits are essential in dairy cattle breeding, and improving body conformation is considered beneficial for reproductive performance. This study systematically analyzed the genetic relationships between six key conformation traits—stature (ST), body depth (BD), loin strength (LS), rump angle (RA), rump width (RW), bone quality (BQ)—and reproductive performance in 1631 Jersey cattle from China. Heritability estimates for conformation traits ranged from 0.05 to 0.62. We identified significant phenotypic and genetic correlations between conformation and reproductive traits, and regression analyses confirmed the predictive value of conformation traits for reproductive outcomes. Genome-wide association studies detected 24 significant SNPs associated with ST, RW, RA, and BQ. Subsequent bioinformatics analysis revealed seven candidate genes (AZIN1, OR2H1, HS6ST3, ERCC4, KCNH5, KRT19, KRT35) involved in embryonic development and estrous cycle regulation. Notably, incorporating six SNPs, which are linked to these candidate genes, into genomic prediction models significantly improved the accuracy for predicting Age at First Calving (AFC) and Gestation Length (GL). These results elucidate the shared genetic basis of conformation and reproduction, providing theoretical support for using conformation traits in marker-assisted selection to enhance reproductive efficiency in Jersey cattle. Full article

Mn⁴⁺-activated fluoride phosphors possess outstanding luminescent properties, making them highly suitable for applications in lighting and display technologies. However, the synthesis of such phosphors generally requires the use of large amounts of highly toxic aqueous HF, leading to serious environmental pollution. To eliminate the use of hazardous HF solution, a low-temperature molten salt method employing NH₄HF₂ was developed to synthesize the narrow-band red emitter Cs₂GeF₆: Mn⁴⁺ phosphor. Following the reaction, the product was washed with a dilute H₂O₂ solution to remove residual NH₄HF₂ and other impurities. The phase purity and morphology were analyzed by X-ray diffraction (XRD) and scanning electron microscopy (SEM), respectively, and the luminescence properties were examined via photoluminescence (PL) spectroscopy. The obtained phosphors exhibit bright red emission characteristics of Mn⁴⁺ under blue-violet excitation. Among them, Cs₂GeF₆: 0.08 Mn⁴⁺ shows the highest emission intensity, with an internal quantum efficiency (IQE) of 78%. A white light-emitting diode (WLED) fabricated by combining this phosphor with a blue chip and commercial Y₃Al₅O₁₂: Ce³⁺ (YAG) phosphor achieved a high luminous efficacy (LE) of ~146 lm/W, a correlated color temperature (CCT) of ~4396 K, and a color rendering index (Ra) of ~83, alongside excellent operational color stability. Full article

Microplastic pollution of ecosystems is considered a modern problem. Freshwater ecosystems, despite the interest shown in their study, remain poorly understood. Lake Baikal (Russia) is one of the least studied freshwater ecosystems in this regard. This large lake is distinguished from others by its high level of biodiversity and clean drinking water. The aim of this study is a multi-matrix investigation of microplastic pollution in one of the lake’s bay. The following matrices are used: surface water, water column, sediment, macrophytes, macroinvertebrates, and fish, as well as ice and snow during the winter. The results show that certain locations exhibit high concentrations of microplastic particles. In some cases, this was due to the properties or characteristics of these locations (littoral zones near the water’s edge, macrophytes with mucus sheaths, ice and snow (potentially, the near-surface water layer after ice melt)), while in others, it was due to localized pollution (pier and ship mooring areas). An analysis of the polymer types of the detected microplastic particles reveals the presence of both common (polypropylene, polyethylene terephthalate, polystyrene, polyethylene, polyvinyl chloride) and rare (polyvinyl alcohol and alkyd resin). Moreover, in some locations, the latter two polymers predominate, a phenomenon rarely observed in other studies. Further research was recommended to focus on the chronic effects of microplastic particles on organisms associated with areas of elevated particle concentrations. Full article

The observed shortage of water resources in the western and southern regions of the Russian Federation may soon affect the territory of the Republic of Bashkortostan. An increase in the share of groundwaters can help to solve this problem. To provide the population of the republic with water resources, the groundwater of magnesium–bicarbonate-type from the Kraka ophiolite massifs can be used. The massifs occur on the western slope of the Southers Urals. In this work we studied ultramafic rocks and their influence on the formation of the chemical composition of water. The research area is located in the northern part of the Zilair synclinorium, which occurs within the Central Ural megazone. In terms of hydrogeology, of particular importance to the territory of the synclinorium is the Zilair basin of fracture waters of the second order, which is part of the Uralian hydrogeologic folded zone. The ultramafic rocks from the studied area have consistently high CaO/Al₂O₃ ratios (0.4–1.6), which indicates the widespread development of parageneses with participation of clinopyroxene and a low degree of depletion of the primitive mantle source. Because of the complex geological structure of the area, water samples collected from both water points in the Kraka massifs, and the surrounding Early–Middle Paleozoic rocks were analyzed for major ions using a laboratory method to identify possible hydro-geochemical zoning. A statistical analysis was then conducted based on the obtained anion–cation composition data. From the viewpoint of the hydrolytic concept, the formation of the chemical composition of groundwater takes place due to the removal of Mg²⁺ from the rock-forming minerals of ultramafic rocks (olivine and pyroxene) and the supply of Na⁺, K⁺, Ca²⁺, and SO₄²⁻ Cl⁻ from atmospheric precipitations. The bicarbonate anion has a complex nature, where both biochemical processes in the soil and atmospheric precipitation play a significant role. Magnesium–bicarbonate-type of waters, due to low mineralization (to 1 g/L) and the majority of other geochemical parameters (pH of the medium, and content of Na, K, Ca, SO₄, and Cl), whose values that are within the limits set by the World Health Organization (WHO), can be used as drinking water. The increased values of total hardness (0.20–3.39 mmol/L) in accordance with the regulatory document SanPiN 1.2.3685–21, adopted by the Russian Federation, do not exceed the maximum permissible concentrations (up to 7.00 (10.00) mEq/L or 3.50 (5.00) mmol/L). The high magnesium content, in accordance with GOST (state standard) R 54316–2020, allows the magnesium–bicarbonate waters of the Kraka massifs to be classified as table mineral waters for the treatment of various diseases (including hypomagnesemia). Full article

Chronic kidney disease (CKD) is a major complication of diabetes, affecting approximately 30–40% of patients. In many cases, CKD progresses to end-stage renal disease (ESRD). The peptide apelin and its receptor, APLNR, which is expressed in the endothelial cells of renal blood vessels, play a key role in glucose uptake and the regulation of vasodilation in the afferent and efferent glomerular arterioles. Numerous studies have demonstrated that the apelinergic system is dysregulated in various pathologies, including CKD in people with diabetes. In recent years, the apelinergic system has emerged as a promising therapeutic target for several diseases, with various apelin analogs and inhibitors being developed. In this review, we summarize the most recent literature on apelin and its cellular mechanisms of action, highlighting the role of the apelinergic system in various pathologies and its impact on patients with CKD and diabetes. Additionally, we explore the currently available analogs and inhibitors and discuss their potential therapeutic applications. Full article

The complexity of field environments poses significant challenges for the segmentation of cotton top leaves, a critical step for apical bud localization in intelligent topping systems. Conventional segmentation models typically rely on large annotated datasets and high computational costs to achieve high precision and robustness. To address these challenges, this paper proposes an efficient and accurate segmentation model, CF-SAM, built upon the Segment Anything Model (SAM) framework. CF-SAM integrates a lightweight Tiny-ViT encoder to reduce computational overhead and employs a LoRA-based fine-tuning strategy for domain adaptation, achieving improved performance with minimal parameter increments. In addition, an Adaptive Prompting Strategy (APS) is introduced to automatically generate high-quality point prompts, enabling fully automated and end-to-end instance segmentation. Trained on only 1000 field images, CF-SAM achieves 98.0% mask accuracy and an mAP@0.5 of 97.83%, while maintaining real-time inference at 58 FPS with only 0.091 M (0.8%) additional parameters. These results demonstrate that CF-SAM achieves an excellent balance between segmentation accuracy and computational cost, providing a reliable technical foundation for apical bud localization and precision agricultural operations. Full article