Search Results (8,843)

Background: Strychnos madagascariensis Poir (Loganiaceae) is a drought-tolerant indigenous fruit tree of East and southern Africa, valued for its food, medicinal, and socio-economic contributions to rural communities. Despite its importance as a candidate food crop, intraspecific morphological and genetic diversity had not previously been characterized, and no simple sequence repeat (SSR) markers had been developed for this species, leaving breeders and conservation planners without the basic diversity baseline needed to prioritize material for domestication. Methods: This study assessed vegetative and reproductive trait variation, variance components, and broad-sense heritability, and SSR-based genetic diversity among 27 morphologically defined S. madagascariensis morphotypes at Bonamanzi Game Reserve, KwaZulu-Natal, South Africa. Three trees were measured per morphotype (81 trees total), over two growing seasons. Genetic diversity was characterized in one representative tree per morphotype using seventeen newly developed SSR loci, the first such markers reported for this species, and analyzed with population structure (STRUCTURE version 2.3.4), PCA, and Nei’s genetic distance. Results: Twenty-seven morphotypes were identified based on leaf colour, shape, hairiness and size, dominated by grey (41%), elongated (59%), less hairy (48%), and medium-sized (>50–90 mm) leaves. Fruit diameter and mass showed the highest inter-morphotype variation (r = 0.949) and also the highest broad-sense heritability (H² = 55.3% and 47.8%, respectively), indicating strong genetic control of these traits and their suitability as targets for selective breeding. Environmental variance exceeded genotypic variance for most traits. A total of 144 alleles were identified across 17 SSR loci (mean 4.24 alleles/locus; mean PIC = 0.31). Population structure gave a preliminary, tentative signal of two genetic clusters (K = 2) with substantial admixture, which we interpret cautiously, given the limited sampling depth. Conclusions: This is the first study to characterize intraspecific morphological variation in S. madagascariensis and the first to develop SSR markers for the species. The results provide a preliminary, single-site framework for conservation genetics and crop improvement that should be validated with larger, multi-site samples. Grey morphotypes GyEvH1, GyEvH2, GyEvH3, GyRlH1 and GyEH2 combined consistent fruiting performance with favourable fruit-trait values and are proposed as priority candidates for further evaluation in domestication and breeding programmes. Full article

Water scarcity in semi-arid regions threatens seed germination and early crop establishment, driving the development of biodegradable Nature-based Solutions to replace synthetic plastic mulches. Porous cellulose membranes were fabricated from rice husk (RH), banana pseudostem (BP), and sugarcane bagasse (SB) by thermo-chemical extraction and high-shear homogenization (n = 5 replicates per membrane type). Membranes were characterized by ATR-FTIR and scanning electron microscopy, confirming removal of non-cellulosic components and biogenic silica preservation in RH, and revealing biomass-dependent porous architectures linked to mechanical and transport behavior. RH produced the most compact fibrillar matrix (compressive strength: 8.16 ± 0.24 MPa; WVT: 170 ± 60 g m⁻² day⁻¹), BP an open interconnected network with superior deformability (9.83 ± 0.25% elongation) and moisture transport (WVT: 400 ± 100 g m⁻² day⁻¹), and SB the highest moisture-retention capacity (215.7 ± 15.8%). Germination assays with Brassica oleracea var. botrytis under water stress showed SB achieved the highest germination rate (90.5 ± 0.99%), confirming that sustained moisture availability governs germination more decisively than transport rate alone. Soil burial tests confirmed biodegradable behavior across all membranes (R² ≥ 0.995; k = 0.043–0.046 day⁻¹). These findings establish a hydrogen-bond-mediated structure–property–function framework for designing biomass-specific cellulose membranes as biodegradable solutions for water-limited agricultural systems. Full article

Background:Scutellaria baicalensis (Scu) extract has been traditionally used in the treatment of stroke-related syndromes, yet its underlying molecular mechanisms, particularly those involving ferroptosis, remain to be fully elucidated. Purpose: This study aims to validate the hypothesis that Scu extract improves cerebral ischemia-reperfusion injury (CIRI) by inhibiting ferroptosis through the PI3K/AKT signaling pathway. Methods: This study employed middle cerebral artery occlusion (MCAO) in male Sprague-Dawley (SD) rats and oxygen–glucose deprivation/reoxygenation (OGD/R) models to evaluate the protective effects of Scu extract against CIRI. Multiple approaches were integrated to elucidate the underlying mechanisms. Furthermore, a range of experimental techniques, including neurological function assessment, TTC staining, histopathological analysis, biochemical assays, qPCR, transmission electron microscopy (TEM), reactive oxygen species (ROS) detection, Western blotting, and immunofluorescence, were used to comprehensively validate its neuroprotective effects. Results: Scu extract significantly improved neurological outcomes and attenuated brain injury in MCAO rats. Proteomic analysis revealed significant enrichment of ferroptosis-related pathways, which was supported by reduced mitochondrial damage, decreased iron accumulation, and restoration of the SLC7A11/GPX4 axis. Subsequently, UPLC/Q-TOF-MS analysis revealed that four major bioactive components were absorbed in MCAO rats. KEGG pathway analysis based on network pharmacology further indicated that the PI3K/AKT signaling pathway is a key regulatory target. Notably, pharmacological inhibition of PI3K with LY294002 markedly abolished the anti-ferroptotic effects of Scu extract, which was further confirmed in vitro. Conclusions: This study demonstrates that Scu extract confers neuroprotection against CIRI in MCAO rats potentially through inhibiting ferroptosis via activation of the PI3K/AKT pathway. Full article

Background/Objectives: Accurate post-mortem interval (PMI) estimation remains challenging in forensic pathology. Although potassium (K⁺) is the most well-validated single biomarker in vitreous humor (VH), multivariate approaches may enhance precision by capturing the complex cascade of post-mortem biochemical changes. This study aimed to develop and validate a multivariate PMI estimation model incorporating three biochemical markers—potassium, ammonium (NH₄⁺), and albumin (ALB)—in vitreous humor using automated clinical chemistry platforms for practical forensic application. Methods: Vitreous humor samples from 38 autopsy cases with documented PMIs (39.5–285 h; mean, 105.5 h) were analyzed for K⁺ (Cobas C8000), NH₄⁺ (Cobas C8000), and ALB (Immage 800 nephelometry). Univariate and multivariate regression analyses were performed, with the residual standard error (RSE) as the primary measure of accuracy. Model validation was conducted by back-calculating PMI in four samples completely distinct from the training cohort. Results: All three analytes demonstrated strong individual correlations with PMI (R²: K⁺ = 0.88, ALB = 0.78, NH₄⁺ = 0.69; all p < 0.001). The multivariate regression model [PMI = 40.25[Alb] + 0.01573[NH₄⁺] + 5.339[K⁺] − 53.032] yielded an RMSE of ±15.5 h (MSE = 240.25 h²), outperforming potassium-only models (RMSE = ±22.6 h). Although NH₄⁺ showed limited statistical significance in the multivariate model (p = 0.128), its inclusion improved overall predictive accuracy. External validation in an independent cohort of four subjects (distinct from the 38 subjects in the training set) demonstrated a mean absolute error (MAE) of 20.4 h. Conclusions: The multivariate approach combining K⁺, NH₄⁺, and ALB in VH improves PMI estimation accuracy compared with single-marker methods. The use of automated clinical chemistry platforms enhances reproducibility and facilitates practical implementation in forensic laboratories. Full article

This research paper focused on the synthesis of 1-[2-hydroxy-3-(phenylcarbamoyloxy)propyl]-4-(R¹, R²-substituted phenyl)piperazin-1-ium chlorides (I)–(III), containing R¹, R² = H, Cl and/or OCH₃, and the evaluation of some of their physicochemical parameters. The in vitro biological investigation of these N-arylpiperazine (NAP) derivatives consisted in assessing their impact on purinergic P2X7-associated signaling, that is, the evaluation of antioxidant, anti-inflammatory and immunomodulatory characteristics. The ultraviolet type C (UVC) irradiation (λ = 254 nm, 0.954 kJ/m²) induced a pronounced stress response in human leukocytes without marked cytotoxicity while maintaining high cell viability (≥90%), as evidenced by increased interleukin (IL)-1β production (94%), elevated IL-1β mRNA expression, enhanced lipid peroxidation (66%), and increased intracellular adenosine 5′-triphosphate (ATP; 97%), respectively. Under basal conditions, these lipophilic NAPs, defined with logarithmic values of retention (capacity) factors corresponding to 100% water in isocratic elution RP-HPLC, i.e., k_w descriptors (varying from 2.3829 to 4.3689), and isocratic chromatographic hydrophobicity index (φ₀) parameters (ranging from 0.7578 to 0.8842), reduced IL-1β production (by 26–63%) and enhanced superoxide dismutase (SOD) activity (up to 64%) without inducing oxidative damage. Under UVC-induced stress, all evaluated compounds decreased lipid peroxidation (up to 45%) and significantly increased antioxidant enzyme activities, including SOD (up to 223%) as well as catalase (up to 145%). The observed effects were associated with changes in intracellular ATP levels and redox-related parameters. In the experiments described in this paper, intracellular ATP was measured so that no direct conclusions could be drawn regarding the extracellular ATP-dependent activation of purinergic receptors, including P2X7. Overall, the results demonstrated that variations within the structure of these NAPs significantly affected compounds’ biological activity, highlighting their potential for further optimization as cytoprotective and anti-inflammatory agents. Full article

Accurate and rapid detection of Volvariella volvacea (straw mushroom) fruiting bodies at harvestable maturity is a critical prerequisite for automated industrial cultivation. However, existing detection methods often yield high false-negative and false-positive rates when processing a small-scale, densely distributed, and heavily occluded targets against complex straw substrate backgrounds. Furthermore, these methods frequently struggle to balance the competing requirements of architectural efficiency (such as parameter volume and computational complexity) and real-time performance for edge computing. To address these challenges, this study proposes a YOLO11n-CPDM, a lightweight detection model based on an improved YOLO11n architecture. The model incorporates synergistic optimizations across feature extraction, fusion, and reconstruction. First, a Dual Coordinate Attention Feature Extraction mechanism is integrated into the C3k2 bottleneck blocks of the backbone network. This enhances target perception in complex, occluded environments by concurrently modeling global context and local salient features. Second, within the neck network, the standard attention module is replaced with the PnPNystraAttention module, coupled with the DySample dynamic upsampling operator. This modification strengthens contextual relationships among multi-scale features and improves spatial consistency during reconstruction while preserving linear computational complexity. Finally, the detection head is optimized using MBConv blocks based on an inverted residual structure to minimize parameter volume. Experimental results on a custom V. volvacea dataset demonstrate that the proposed YOLO11n-CPDM model achieves significant performance gains, with Precision (P), Recall (R), and Mean Average Precision (mAP50) reaching 86.8%, 87.5%, and 88.4%, respectively. These figures represent improvements of 2.7, 3.0, and 3.2 percentage points over the baseline YOLO11n model. Additionally, the model size is reduced to 4.8 MB (a 12.7% decrease), while achieving inference speeds of 42.7 FPS on Jetson AGX Orin and 21.2 FPS on Jetson Nano, outperforming the baseline model on both embedded platforms. Consequently, the proposed model effectively enhances detection performance in complex environments while maintaining excellent lightweight characteristics and deployment flexibility, providing a solid technical foundation for intelligent perception and automated harvesting of V. volvacea. Full article

For hydrological parameterization in high-Andean catchments, it is necessary to understand whether near-surface hydro-structural soil properties can provide a surrogate signal of particle-size composition when direct texture information is sparse. This study evaluated the extent to which sand, silt, and clay fractions can be approximated from organic matter ($\mathrm{OM}$), bulk density (${\rho }_b$), and saturated hydraulic conductivity ($K_{\mathrm{sat}}$) in the Zamora Huayco (ZH) and Irquis catchments, southern Ecuador. A harmonized dataset ($n=44$) was analyzed through exploratory statistics, compositional assessment, correlation analysis, PCA, fraction-wise regression, $\mathrm{ILR}$-based modeling, AIC/BIC term reduction, sensitivity analysis excluding $\mathrm{OM}$, nested $\mathrm{LOOCV}$, and bootstrap-based uncertainty intervals. Among LULC classes, samples classified as paramo occupied a distinct high-Andean hydro-edaphic domain, characterized by a differentiated relationship between soil physical properties and hydrological behavior. PCA showed that the dominant covariance structure involved $\mathrm{OM}$, ${\rho }_b$, $K_{\mathrm{sat}}$, and the redistribution between sand and silt. The BIC-reduced $\mathrm{ILR}$ model provided the most balanced formulation, with positive nested $\mathrm{LOOCV}$ performance for sand, silt, and clay ($R_{\mathrm{LOOCV}}^2=0.147$, $0.704$, and $0.124$, respectively) and exact $100\%$ compositional closure after inverse transformation. Silt was the most stable predicted fraction, whereas sand and clay retained larger residual uncertainty, stronger tail departures, and partial compression of the observed variability. The proposed equations provide local hydro-pedotransfer support, although their predictive signal remains dependent on further refinement, uncertainty assessment, and external validation before regional application. Full article

This study investigated the impact of recreational use on trails in the Atlantic Forest (Ubatuba Municipality, São Paulo State, Brazil) using physical, chemical and geochemical indicators. Five trails with different morphological characteristics were selected, and paired samples were collected from the trail surface (TR) and trail-side slope (TA). The statistical approach combined local analyses for each trail with global clustering (n = 19) using Student’s t-test, along with multivariate modeling through Principal Component Analysis (PCA) and Pearson correlation. The analysis included physical attributes (bulk density, particle size and porosity), chemical attributes (pH, organic matter and macronutrients) and geochemical compositions (major oxides and trace elements determined by XRF). The overall results reveal systematic compaction in the trail surface (TR), with bulk density increasing from 1.32 g/cm³ (TA) to 1.37 g/cm³ (TR) (p = 0.038), and total porosity decreasing from 47.26% to 45.34% (p = 0.016). In contrast, the geochemical oxide composition (SiO₂, Al₂O₃, Fe₂O₃) remained stable (p > 0.05), indicating the resilience of the mineral matrix. However, significant local dynamics (p < 0.05) in K₂O and MgO were observed in more preserved trails, associated with surface compaction and fragmentation of the litter layer, and phosphorus showed strong dependence on organic matter (r = 0.85). Multivariate analysis indicates that degradation is predominantly physical and micromorphological at the local scale, with bulk density and porosity being the most sensitive indicators for environmental monitoring. Full article

The constitutive analysis model and hot processing map of the ZM6 alloy across various deformation conditions were investigated during hot compression experiments. True stress-strain curves within 300–450 °C and 0.0001–0.1 s⁻¹ were obtained from compression tests on a Gleeble-1500 platform. The results showed that higher strain rates (e.g., 0.1 s⁻¹) induced pronounced work hardening, whereas high temperatures (300–400 °C) combined with low strain rates (10⁻⁴ s⁻¹) promoted conditions conducive to dynamic recrystallization (DRX), leading to a softening tendency of steady-state flow stress. Additionally, a modified strain-compensated constitutive model was built for flow stress prediction. Material constants were plotted as fifth-order polynomial functions of strain (0.025–0.80) for precise stress predictions. The derived activation energy (Q = 182.38 kJ/mol) falls within the typical range for Mg-RE alloys. Leave-one-temperature-out cross-validation showed average AARE values of 7.2–9.8%, demonstrating the model’s interpolation capability and its sensitivity to extrapolation. Cross-validation within the training dataset showed reasonable consistency between experimental and predicted stresses (R > 0.997, AARE < 4.35%). Using the dynamic materials model, hot processing maps identified safe deformation zones and instability zones of the ZM6 alloy. Flow instability was observed at strain rates >0.01 s⁻¹, particularly at low temperatures (300–350 °C). Optimal processing windows appeared in high-energy dissipation (η > 30%) regions, e.g., 400–450 °C/10⁻⁴–10⁻³ s⁻¹. Optical microscopy confirmed that at high temperatures (≥400 °C) and low strain rates (≤0.001 s⁻¹), a uniform, fine-grained, fully recrystallized structure can be obtained, whereas low temperatures (350 °C) and high strain rates (0.1 s⁻¹) produce coarse elongated grains with limited DRX, consistent with the instability regime predicted by the processing maps. Under intermediate conditions (e.g., 400 °C, 0.01 s⁻¹), a bimodal grain distribution indicates incomplete recrystallization. Although EBSD analysis was not performed in this study, the optical microstructures directly validate the predicted safe and unstable windows. Together, all these findings provide preliminary model-based guidance for optimizing hot working parameters to balance microstructural stability and processing efficiency. Full article

Small pelagic fishes are essential for artisanal fisheries and food security in São Tomé and Príncipe, yet biological information required for stock assessment remains scarce. This study examined the population dynamics and life-history traits of Caranx crysos, Euthynnus alletteratus, Hemiramphus balao, and Cheilopogon melanurus using 9052 specimens collected from artisanal landings between 2023 and 2025. C. melanurus (35.2%) and H. balao (34.0%) dominated the sampled catches, followed by C. crysos (18.1%) and E. alletteratus (12.7%). Standardized CPUE series modelled using GAMs revealed significant temporal and seasonal variation in relative abundance, with contrasting species-specific trends. Length–weight relationships revealed negative allometric growth in three of the four species examined (75%), with b values ranging from 2.50 to 3.19, while Fulton’s condition factor differed significantly among species (Kruskal–Wallis χ² = 6702.7, p < 0.001). Sex-ratio analyses showed significant deviations from parity in C. crysos and C. melanurus, whereas E. alletteratus and H. balao remained balanced. Gonadosomatic index and maturity-stage distributions indicated year-round reproductive activity with distinct spawning peaks. Length at first maturity (L50) ranged from 30.2 cm to 38.8 cm among species. Growth parameters estimated from length-frequency data using the von Bertalanffy Growth Function fitted through ELEFAN_GA in TropFishR yielded L∞ values of 43.9–68.4 cm and K values of 0.065–0.336 yr⁻¹. Growth performance index (φ′) ranged from 2.48 to 2.99, corresponding to theoretical longevities of 8.9–46.3 years. Length-based cohort analysis indicated biomass concentration in intermediate size classes and increasing fishing mortality towards larger individuals. Exploitation rates revealed contrasting fishing pressures among species, while sensitivity analyses identified growth and mortality parameters as the main sources of uncertainty. These findings provide the first integrated biological baseline for the assessment and management of small pelagic resources in São Tomé and Príncipe. Full article

Accurate short-term forecasting for energy consumption is crucial in modern energy network management, especially for cities such as Tetouan, where considerable climate variability and diverse usage patterns present significant challenges when it comes to making short-term forecasts. This paper proposes a hybrid residual learning framework that combines a long short-term memory (LSTM) network with eXtreme Gradient Boosting (XGBoost) to improve short-term load forecasting for the Tetouan electricity network. The novelty of the proposed approach lies in coupling temporal sequence modeling with residual error correction driven by exogenous meteorological and calendar-related information. The proposed model is validated using real electricity consumption data from Zone 2 of Tetouan City, with further validation across all three available zones confirming the model’s generalizability. The proposed model achieves a coefficient of determination (R²) of 0.984, an RMSE of 687.21 kWh, and a MAPE of 2.41%, representing a 121.3 kWh RMSE improvement over the standalone LSTM baseline. These results confirm that the hybrid model is better at tracking periods of high demand compared to conventional machine learning approaches and standalone deep learning models. Full article

Background: Obstructive sleep apnea (OSA) is characterized by chronic intermittent hypoxia (CIH), which causes numerous metabolic changes, leading to non-alcoholic fatty liver disease (NAFLD). Our study explored the suggested role of hypoxia-inducible factor 1-α (HIF-1α) in the pathophysiological mechanisms linking OSA with NAFLD. Methods: This case-control study was conducted at the Sleep Disorders Unit at Qena University Hospital from March 2022 to October 2023, including 64 subjects (48 OSA patients; in a secondary analysis, OSA patients were further stratified according to the presence or absence of NAFLD–16 controls) who were subjected to a polysomnography (PSG) for apnea hypopnea index (AHI) and transient elastography for controlled attenuation parameter (CAP) score and liver stiffness measurement (LSM). Serum levels of HIF 1-α, fasting blood glucose, and fasting insulin were measured. Results: HIF-1α level showed the highest significant value was in the severe group (p = 0.001). Additionally, the severe group had the highest LSM compared to the other groups (p = 0.032). OSA patients with NAFLD, compared to OSA patients without NAFLD, showed significantly higher BMI (42.74 vs. 29.11 kg/m², p < 0.001), homeostatic model assessment for insulin resistance (HOMA-IR) mean score (3.92 vs. 1.21, p < 0.0001), HIF-1α level (6.01 vs. 2.14 ng/L, p = 0.045), and the LSM score (5.55 vs. 3.85 kPa, p < 0.001). HIF-1α showed significant positive correlations with AHI (r = 0.515, p < 0.001), waist circumference WC (r = 0.291, p = 0.045), HSI (r = 0.3, p = 0.038), and CAP score (r = 0.288, p = 0.047). Conclusions: Although serum HIF-1α levels were significantly higher in OSA patients with NAFLD and correlated with indices of hepatic steatosis, HIF-1α was not identified as an independent predictor of NAFLD after adjustment for metabolic confounders, suggesting a potential role of hypoxia-responsive pathways in pathophysiology of NAFLD in OSA. Full article

Adenoviral vectors are widely used for gene therapy and vaccine development. To circumvent pre-existing immunity against commonly used human adenovirus type 5, vectors based on rare human serotype or animal adenoviruses have attracted increasing interest. Previously, we constructed vectors based on fowl adenovirus 4 (FAdV-4) and replaced the knob of FAdV-4 fiber2 with that of FAdV-1 fiber1 to generate FAdV4-CF1K vectors with enhanced transduction efficiency in human cells. In this study, we aimed to modify the packaging system to efficiently produce FAdV-4 vectors carrying transgenes toxic to viral replication. Chicken LMH cells failed to form colonies at low seeding densities. We collected used medium from LMH cell cultures and used it as a supplement to adapt LMH cells, generating the colony-competent subclone LMH-C3532. A lentiviral vector encoding a codon-optimized tetracycline repressor (tetR) was transduced into LMH-C3532 to establish a tetR-integrated cell line, LMH-tetR24. An adenoviral plasmid, pKFAV4-CF1K-CtG, was constructed in which a tetracycline operator (tetO)-bearing CMV promoter controlled GFP expression. The SwaI-flanked GFP in this plasmid was replaced with the HA gene from an H5N1 influenza virus to generate pKFAV4-CF1K-CtHA. Linearized adenoviral plasmids were transfected into LMH-tetR24 cells, and recombinant FAdV4-CF1K-CtG and FAdV4-CF1K-CtHA viruses were successfully rescued, amplified, and purified. When infected with FAdV4-CF1K-CtG at various multiplicities of infection (MOI), the progeny virus yield from LMH-tetR24 cells was 4–10 times higher than that from LMH-C3532 cells. For FAdV4-CF1K-CtHA, the yield difference between the two cell lines was even more pronounced, reaching 3–4 orders of magnitude. Overexpression of HA in LMH-C3532 cells negatively affected FAdV4-CF1K-CtHA replication, resulting in smaller and fewer plaques. In conclusion, by separately integrating tetR into packaging cells and TetO into the adenoviral plasmid, we established a system that can be routinely used to package FAdV-4 vectors. Notably, this system facilitates the propagation of FAdV-4 vectors carrying toxic transgenes. Full article

Cooking liquid from Manila clam (Ruditapes philippinarum) is an underutilized byproduct rich in water-soluble taste compounds, representing a potential source of natural umami peptides. In this study, peptide fractions were separated from the cooking liquid. A total of 764 peptide sequences were identified from the most potent fraction, F3 (<3 kDa), by UPLC-ESI-Q-TOF-MS/MS. Machine learning prediction and molecular docking were further used for screening. Five candidate peptides were selected: TQDTVVALDA, KEY, YKD, RND, and GEAF. Sensory evaluation (on a 0–5 scale) and electronic tongue measurements independently confirmed that peptide YKD possessed the strongest taste profile, with an electronic tongue relative umami score of 8.81 ± 0.22. Furthermore, cell-based assays demonstrated that YKD effectively up-regulated the transcriptional expression of taste-related receptors, including GPRC6A, in STC-1 cells, revealing a multi-receptor synergetic mechanism for umami perception. In STC-1 cells, all peptides induced intracellular Ca²⁺ responses and showed no obvious cytotoxicity at 0.5–8.0 mmol/L. YKD produced the highest fluorescence response (0.59) at 1.0 mmol/L. Quantitative RT-PCR analysis suggested that YKD was associated with T1R1/T1R3-related expression, whereas TQDTVVALDA induced stronger CaSR expression. These findings elucidate the specific peptide sequence that engages multiple receptors to create complex tastes, providing a theoretical basis for converting seafood processing byproducts into natural flavor enhancers. Full article

To address the insufficient accuracy of short-term load forecasting caused by the strong randomness of distributed energy output, variable electricity consumption patterns, and complex meteorological factors, this study proposes a load forecasting method that integrates K-means scene clustering and a Transformer–BiGRU–Attention (CTBA) hybrid deep learning architecture. Different from conventional Transformer–BiGRU hybrid forecasters that train a single global predictor across all operating conditions, the proposed CTBA framework first partitions daily load curves into representative scenes and then routes each sample to a scene-specific Transformer–BiGRU–Attention predictor, thereby reducing distributional heterogeneity before temporal modeling. First, the K-means algorithm is used to perform scene clustering on historical daily load curves, and the optimal number of clusters is selected according to the silhouette coefficient and downstream prediction performance. Subsequently, the CTBA model is trained separately for each clustering subset. The Transformer encoder captures the long-range global dependencies of load sequences through the self-attention mechanism, the BiGRU module extracts local bidirectional temporal fluctuation features, and the Attention mechanism further focuses on key time nodes such as morning and evening peaks while fusing multi-source data including historical load, day-ahead electricity price, and multi-dimensional meteorological factors. Experimental results based on the German ENTSO-E power dataset show that the coefficient of determination R² of the proposed model reaches 0.9893, with MAE, RMSE, and MAPE as low as 0.0141, 0.0187, and 3.92%, respectively, which are significantly improved compared to benchmark models such as SVR, LSTM, CNN, and TCN-BiGRU. Ablation experiments further demonstrate that removing the clustering, Transformer, BiGRU, or attention layer will degrade performance, thus verifying the effectiveness and superiority of the method in short-term load forecasting and providing an accurate solution for the short-term load forecasting of power systems. Full article