Search Results (727)

The year 2025 has been transformative in breast oncology, marked by the maturation of pivotal adjuvant trials, the introduction of novel ADCs, and the validation of proactive biomarker-driven strategies across all molecular subtypes. ASCO, ESMO, and SABCS contributed pivotal updates that further refined treatment paradigms. This systematic review synthesizes and critically evaluates pivotal Phase II/III clinical trials presented at major oncology conferences (ASCO 2025, ESMO 2025, SABCS 2025) and published in high-impact journals during 2025. A curated selection of pivotal Phase II/III trials, and major prospective trials published or presented in 2025 was performed. Data extraction focused on trial design, population, interventions, efficacy endpoints, and safety outcomes. Narrative synthesis was organized by disease stage and molecular subtype. Key 2025 findings (50 clinical trials) include: (1) confirmation of overall survival benefit with adjuvant CDK4/6 inhibitors in HR+/HER2− early breast cancer (monarchE: HR = 0.842, p = 0.0273); (2) establishment of trastuzumab deruxtecan (T-DXd) as a new standard in high-risk HER2+ early disease (DESTINY-Breast05: IDFS HR = 0.47) and first-line metastatic settings (DESTINY-Breast09: PFS HR = 0.58); (3) validation of TROP2-directed ADCs as first-line therapy for metastatic triple-negative breast cancer (ASCENT-03: PFS HR = 0.62; BEGONIA: ORR 79%); (4) paradigm shift to proactive, liquid biopsy-guided therapy switching (SERENA-6: PFS HR = 0.44); (5) updated efficacy and safety of the oral SERD imlunestrant from the EMBER-3 trial, supporting its role in ESR1-mutated advanced breast cancer and in combination with abemaciclib; (6) confirmation of long-term survival benefit for neoadjuvant carboplatin in early TNBC and new positive adjuvant data; (7) pivotal advances in HER2+ metastatic disease sequencing with tucatinib and T-DXd; (8) evidence supporting optimized adjuvant endocrine therapy in HER2+/HR+ early disease; and (9) emergence of novel agents with improved therapeutic indices, including PROTAC degraders, oral SERDs, and mutant-selective PI3K inhibitors. The 2025 evidence base has fundamentally reshaped breast cancer management, establishing new standards of care across all subtypes. Unifying themes include biomarker-driven personalization, strategic treatment sequencing, management of unique toxicities, and emphasis on patient-reported outcomes. Future challenges include optimizing treatment integration, managing financial toxicity, and ensuring equitable global access. Full article

Extruded polypropylene (PP) films were exposed to cold air plasma treatment, which resulted in significant changes in their bulk properties. The maximal elongation, ultimate tensile strength (UTS), and toughness of the films were increased. The toughness of the films increased from $U_{T0}=(3323\pm 400) \mathrm{M}\mathrm{P}\mathrm{a}$ to $U_{T\_PT}=(4434\pm 400) \mathrm{M}\mathrm{P}\mathrm{a}$, which is due to the growth of both the maximal elongation and the UTS of the plasma-treated samples. We relate the improvement of the mechanical properties of PP to the morphological transformations revealed in the plasma-treated PP films. Plasma treatment of PP samples was also followed by the modification of their surface properties. Plasma treatment resulted in hydrophilization of PP films followed by hydrophobic recovery. The bulk and surface properties of the plasma-treated PP films evolve with time. The following hierarchy of the temporal scales related to the studied relaxation processes is established: ${\tau }_{HR}>{\tau }_{\epsilon }={\tau }_T={\tau }_{UTS}>{\tau }_E$, where ${\tau }_{HR}, {\tau }_{\epsilon }, {\tau }_T, {\tau }_{UTS}$ and ${\tau }_E$ are the time scales of the change in the apparent contact angle (hydrophobic recovery), elongation, toughness, ultimate tensile strength, and Young modulus, respectively. The longest of the relaxation times is related to the surface processes, i.e., hydrophobic recovery. The stress–strain curves of the untreated virgin and plasma-treated PP are well described with the twin-slope linear dependencies. The post-plasma-treatment recovery of the tangent modulus is reported. Cold plasma treatment of polypropylene produces surface oxidation and functionalization, evidenced by the emergence of C–O, C=O, and COOH functionalities. Full article

In complex natural scenarios, facial expression recognition often encounters partial occlusions caused by glasses, hand gestures, and hairstyles, making it difficult for models to extract effective features and thereby reducing recognition accuracy. Existing methods often employ attention mechanisms to enhance expression-related features, but they fail to adequately address the issue where high-frequency responses in occluded regions can disperse attention weights (e.g., incorrectly focus on occluded areas), making it challenging to effectively utilize local cues around the occlusions and limiting performance improvement. To address this, this paper proposes a network based on an adaptive attention mechanism (Adaptive Attention Vision Transformer, AA-ViT). First, an Adaptive Attention module (ADA) is designed to dynamically adjust attention scores in occluded regions, enhancing the effective information in features. Next, a Dual-Branch Multi-Layer Perceptron (DB-MLP) replaces the single linear layer to improve feature representation and model classification capability. Additionally, a Random Erasure (RE) strategy is introduced to enhance model robustness. Finally, to address the issue of model training instability caused by class imbalance in the training dataset, a hybrid loss function combining Focal Loss and Cross-Entropy Loss is adopted to ensure training stability. Experimental results show that AA-ViT achieves expression recognition accuracies of 90.66% and 90.01% on the RAF-DB and FERPlus datasets, respectively, representing improvements of 4.58 and 18.9 percentage points over the baseline ViT model, with only a 24.3% increase in parameter count. Compared to existing methods, the proposed approach demonstrates superior performance in occluded facial expression recognition tasks. Full article

This study examines how technological innovation and sustainability jointly reshape contemporary public administration by integrating digital transformation with public value creation. Using a mixed-method approach, we compile a Scopus-based bibliographic dataset and conduct descriptive and network analyses on 199 articles to map publication trends, methodological patterns, and core keyword clusters. We then perform an in-depth qualitative content analysis of 83 papers, coding public sector domains, actors, technological innovations, and sustainability dimensions. Findings highlight a shift from early e-government, centered on administrative efficiency, toward a paradigm of “sustainable digital governance”, where AI, IoT, blockchain and data analytics drive the twin digital–green transition. Five conceptual clusters and several application domains show that public value increasingly emerges within collaborative ecosystems involving administrations, firms, universities, citizens and digital platforms. The study offers an integrated overview of this evolving field and clarifies technology’s role as an enabling factor in sustainable governance. Building on the review results, we propose the Sustainable Public Innovation Ecosystem (SPIE) framework, which links systemic enablers (technological and sustainability innovation) governance efficiency and sustainable public value through ecosystem dynamics and governance mechanisms. It also outlines a future research agenda on hybrid actors ethical and regulatory issues, and approaches to measuring sustainable public value, providing guidance for scholars and policymakers designing digitally enabled and sustainability-oriented public reforms. Full article

The H2IOSC project aims to establish a federated cluster of European distributed research infrastructures involved in the humanities and cultural heritage sectors, with operating nodes across Italy. Through four key RIs—DARIAH-IT, CLARIN, OPERAS, and E-RIHS—the project promotes collaboration among researchers with interdisciplinary expertise. Within this framework, DIGILAB functions as the digital access platform for the Italian node of E-RIHS. Conceived as a socio-technical infrastructure for the Heritage Science community, DIGILAB is designed to manage heterogeneous data and metadata through advanced knowledge graph representations. The platform adheres to the FAIR principles and supports the complete data lifecycle, enabling the development and maintenance of Heritage Digital Twins. DIGILAB integrates diverse categories of information related to cultural sites and objects, encompassing historical and artistic datasets, diagnostic analyses, 3D models, and real-time monitoring data. This monitoring capability is achieved through the deployment of cutting-edge Internet of Things (IoT) technologies and large-scale Wireless Sensor Networks (WSNs). As part of DIGILAB, we developed SENNSE (v1.0), a fully open hardware/software platform dedicated to environmental and structural monitoring. SENNSE allows the remote, real-time observation and control of cultural heritage sites (collecting microclimatic parameters such as temperature, humidity, noise levels) and of cultural objects (collecting object-specific data including vibrations, light intensity, and ultraviolet radiation). The visualization and analytical tools integrated within SENNSE transform these datasets into actionable insights, thereby supporting advanced research and conservation strategies within the Cultural Heritage domain. In the following sections, we provide a detailed description of the SENNSE platform, outlining its hardware components and software modules, and discussing its benefits. Furthermore, we illustrate its application through two representative use cases: one conducted in a controlled laboratory environment and another implemented in a real-world heritage context, exemplified by the “Biblioteca Bernardini” in Lecce, Italy. Full article

In recent years, Transformer-based methods have demonstrated proficiency in capturing complex patterns for time series forecasting. However, their quadratic complexity relative to input sequence length poses a significant bottleneck for scalability and real-world deployment. Recently, the Mamba architecture has emerged as a compelling alternative by mitigating the prohibitive computational overhead and latency inherent in Transformers. Nevertheless, a vanilla Mamba backbone often struggles to adequately characterize intricate temporal dynamics, particularly long-term trend shifts and non-stationary behaviors. To bridge the gap between Mamba’s global scanning and local dependency modeling, we propose C-T-Mamba, a hybrid framework that synergistically integrates a Mamba block, channel attention, and a temporal convolution block. Specifically, the Mamba block is leveraged to capture long-range temporal dependencies with linear scaling, the channel attention mechanism filters redundant information, and the temporal convolution block extracts multi-scale local and global features. Extensive experiments on five public benchmarks demonstrate that C-T-Mamba consistently outperforms state-of-the-art (SOTA) baselines (e.g., PatchTST and iTransformer), achieving average reductions of 4.3–18.5% in MSE and 3.9–16.2% in MAE compared to representative Transformer-based and CNN-based models. Inference scaling analysis reveals that C-T-Mamba effectively breaks the computational bottleneck; at a horizon of 1536, it achieves an $8.8\times$ reduction in GPU memory and over 10× speedup compared to standard Transformers. At 2048 steps, its latency remains as low as 8.9 ms, demonstrating superior linear scaling. These results underscore that C-T-Mamba achieves SOTA accuracy while maintaining a minimal computational footprint, making it highly effective for long-term multivariate time series forecasting. Full article

CRISPR-Cas9 technology has opened new perspectives in genome editing of clonally, asexually propagated and polyploid plants by enabling multiple allelic gene edits. Traditional Agrobacterium- and particle bombardment-mediated transformations, which rely on integration of gene-editing transgene cassettes, have been efficiently applied to several plants; however, concerns about the acceptability of resultant edited transgenic genotypes make these methods less attractive for vegetatively propagated crops. We leveraged and optimized the CRISPR-Cas9/sgRNA-RNPs system for delivery into protoplasts of the hexaploid sweetpotato cultivar PI-318846, targeting eukaryotic translation initiation factor isoform 4E genes to enhance resistance to SPFMV potyviruses. To evaluate the efficiency of pre-assembled Cas9/sgRNA-RNP in sweetpotato transfection, single guide RNAs were designed to target putative host susceptibility genes: IbeIF4E, IbeIF(iso)4E, and IbCBP. Freshly isolated leaf protoplasts were subjected to CRISPR-CAS9-RNP PEG-mediated transfection under different parameters. Sweetpotato regenerants screened using PCR-RE-T7 assay, sequencing, and Inference CRISPR Edit analyses of target-site amplicons revealed the most efficient editing conditions utilizing 25% PEG with a 3:1 (15 µg:45 µg) ratio of Cas9/sgRNA-RNP for 25 min and 48 h incubation period. Different allelic InDels were obtained with editing efficiencies of 10–20% in regenerated plantlets, demonstrating that PEG-mediated CRISPR-RNP transfection system is key for advancing DNA-free editing tools in polyploid and vegetatively propagated crops. Full article

The deployment of 5G (Fifth-Generation) networks in industrial Internet of Things (IoT), intelligent transportation, and emergency communications introduces heterogeneous and dynamic network states, leading to frequent and diverse faults. Traditional fault detection methods typically emphasize either local temporal anomalies or global distributional characteristics, but rarely achieve an effective balance between the two. In this paper, we propose a parallel multi-branch convolutional neural network (CNN)–Transformer framework (MBCT) to improve fault diagnosis accuracy in 5G networks. Specifically, MBCT takes time-series network key performance indicator (KPI) data as input for training and performs feature extraction through three parallel branches: a CNN branch for local patterns and short-term fluctuations, a Transformer encoder branch for cross-layer and long-term dependencies, and a statistical branch for global features describing quality-of-experience (QoE) metrics. A gating mechanism and feature-weighted fusion are applied outside the branches to adjust inter-branch weights and intra-branch feature sensitivity. The fused representation is then nonlinearly mapped and fed into a classifier to generate the fault category. This paper evaluates the performance of the proposed model on both the publicly available TelecomTS multi-modal 5G network observability dataset and a self-collected SDR5GFD dataset based on software-defined radio (SDR). Experimental results demonstrate that the proposed model achieves superior performance in fault classification, achieving 87.7% accuracy on the TelecomTS dataset and 86.3% on the SDR5GFD dataset, outperforming the baseline models CNN, Transformer, and Random Forest. Moreover, the model contains approximately 0.57M parameters and requires about 0.3 MFLOPs per sample for inference, making it suitable for large-scale online fault diagnosis. Full article

Smart campuses employ advanced digital technologies and intelligent communication systems to enhance educational, operational, and living environments. This study investigates stakeholder perceptions of smart campus priorities in Saudi Arabia through a structured questionnaire administered to students and faculty. The study considered King Fahd University of Petroleum and Minerals (KFUPM) in Dhahran as a case study in this regard. The survey examined 22 smart campus aspects grouped into six domains: smart education, smart mobility, smart energy and waste management, smart buildings and work environment, smart safety and security, and smart open spaces. The results indicated strong consensus regarding the importance of all domains, with an overall mean rating of 4.3 out of 5.0 and Relative Importance Index (RII) values ranging from 0.77 to 0.91. The highest-ranked aspects included IoT-enabled cooling energy optimization, smart public transportation, smart lighting systems, smart workflow management, e-libraries, and fire prevention and detection systems, reflecting a pronounced emphasis on infrastructure quality, energy efficiency, and operational effectiveness. The findings suggest that smart campus development in Saudi Arabia should prioritize high-impact, user-valued initiatives that align with Vision 2030 objectives including digital transformation. Strategic early investments in smart buildings, energy management, and mobility systems can deliver measurable benefits in this regard. Further research is recommended to consider additional case studies in the Saudi context to ensure that smart campuses remain contextualized and responsive to user needs. Full article

Rabies virus (RABV) causes fatal encephalitis once it invades the central nervous system (CNS), and treatment options are extremely limited at this stage. We investigated the recombinant RABV variants SPBN, SPBNGA (glycoprotein substitution R333E), SPBNGAK (R333E plus N194K), SPBNGAS (R333E plus N194S), and TriGAS (three copies of the R333E/N194S glycoprotein). We evaluated their cellular tropism and immune activation in an intracerebral mouse infection model using immunohistochemistry and confocal immunofluorescence. SPBNGAK (R333E/N194K) resulted in mixed neuronal and astrocytic infection and lethal disease. In contrast, the R333E/N194S mutations in the GAS variants were associated with reduced neuronal infection and apparent astrocyte-restricted infection patterns. This tropism shift coincided with microglial activation (allograft inflammatory factor 1, amoeboid transformation) and astrocytic activation (nestin), along with T-cell infiltration and endothelial activation that persisted beyond viral clearance. SPBNGAK-infected astrocytes expressed nestin, while GAS variant-infected astrocytes remained nestin-negative and were rapidly cleared. Intracerebral co-inoculation of astrocytotropic TriGAS with the lethal neurotropic DOG4 strain was associated with survival and a marked reduction in detectable DOG4 neuronal infection. These findings suggest that glycoprotein-mediated astrocyte tropism may be associated with altered immune responses after rabies CNS invasion. While mechanistic causality cannot be inferred, these observations may inform the design of future studies exploring astrocyte-restricted RABV infection in therapeutic-related contexts. Full article

Rising global electricity demand and the expansion of distribution networks require a critical assessment of component-level greenhouse gas contributions. Distribution transformers, although indispensable, have significant life-cycle carbon impacts due to the use of materials, manufacturing, and in-service losses. This study conducts a life-cycle assessment of a single-phase, 75 kVA oil-immersed distribution transformer manufactured in Newfoundland, one of the provinces with the cleanest, hydro-dominated grids in Canada, and evaluates it over a 40-year lifespan. Using a cradle-to-use boundary, the analysis quantifies embodied emissions from raw material extraction, manufacturing, and transportation, alongside operational emissions derived from empirically measured no-load and load losses. All the data are collected directly during the manufacturing process, ensuring high analytical fidelity. The energy efficiency of the transformer is analyzed in MATLAB version R2023b using measured no-load and load losses to generate efficiency, load characteristics under various operating conditions. Under varying load factor scenarios and based on Newfoundland’s 2025 grid intensity of 18 g CO₂e/kWh, the lifetime operational emissions are estimated to range from 0.19 t CO₂e under no-load operation to 4.4 t CO₂e under full-load conditions. A linear regression-based decarbonization model using Microsoft Excel projects grid intensity to reach net-zero around 2037, two years beyond the provincial target, indicating that post-2037 transformer losses will remain energetically relevant but carbon-neutral. Sensitivity analysis reveals that temporary overloading can substantially elevate lifetime emissions, emphasizing the value of smart-grid-enabled load management and optimal transformer sizing. Comparative assessment with fossil fuel-intensive provinces across Canada demonstrates the dominant influence of grid generation mix on life-cycle emissions. Additionally, refurbishment scenarios indicate up to 50% reduction in cradle-to-gate emissions through material reuse and oil reclamation. The findings establish a scalable framework for integrating grid decarbonization trajectories, life-cycle carbon modelling, and circular-economy strategies into sustainable distribution network planning and transformer asset management. Full article

The expanding insect farming industry generates up to 67,000 tons of frass per year. Its potential use as fertilizer is promising, but has not yet been widely studied. This study aimed to characterize the chemical composition, organic matter structure, ecotoxicity, and phytotoxicity of frass from four insect species in order to evaluate its potential as a fertilizer. We compared four types of insect frass (IF) (Tenebrio molitor, Galleria mellonella, Hermetia illucens, and Acheta domesticus) to Eisenia fetida vermicompost (EFV). We used physicochemical analyses (pH, electrical conductivity (EC), macro-micronutrients and dissolved organic carbon (DOC), spectroscopy (solid-state ¹³C nuclear magnetic resonance (NMR), and Fourier-transform infrared spectroscopy (FTIR)) and thermogravimetry/differential scanning calorimetry (TGA/DSC: R₁, R₂, Tmax), together with phytotoxicity (germination index, %GI) and ecotoxicity (toxicity units, TU) bioassays. Composition was species-dependent: A. domesticus showed the highest levels of nitrogen (N), phosphorus (P), and potassium (K); the concentration of DOC was higher in insect frass (IF) than in EFV, with the highest concentration found in IF of T. molitor. ¹³C NMR/FTIR profiles distinguished between frass (carbohydrates/proteins and chitin signals) and EFV (humified, oxidized matrix). Thermal stability followed: G. mellonella (R₁ ≈ 0.88) ≥ A. domesticus (0.79) > H. illucens (0.73) > EFV (0.67) > T. molitor (0.50). In bioassays, T. molitor and A. domesticus exhibited phytotoxicity (%GI < 30), whereas G. mellonella and H. illucens did not. EFV exhibited the highest %GI. Dilution increased %GI in all materials, especially in T. molitor and A. domesticus, and reduced acute risk (TU). Frass is not a uniform input: its agronomic performance emerges from the interaction between EC (ionic stress), the availability of labile C (DOC, C/N and low-temperature exotherms), and structural stability (R₁/R₂ and aromaticity). In terms of formulation, IF can provide nutrients that mineralize rapidly, whereas EFV contributes stability. Controlling the inclusion and dilution of materials (e.g., limiting the amount of T. molitor in blends) and considering the mixing matrix helps to manage phytotoxicity and ecotoxicity, and realize the fertilizer value of the product. Full article

In the era of the digital food economy, high-fidelity food images are critical for applications ranging from visual e-commerce presentation to automated dietary assessment. However, developing robust computer vision systems for food analysis is often hindered by data scarcity for long-tail or regional dishes. To address this challenge, we propose a novel high-fidelity food image synthesis framework as an effective data augmentation tool. Unlike generic generative models, our method introduces an Ingredient-Aware Diffusion Model based on the Masked Diffusion Transformer (MaskDiT) architecture. Specifically, we design a Label and Ingredients Encoding (LIE) module and a Cross-Attention (CA) mechanism to explicitly model the relationship between food composition and visual appearance, simulating the “cooking” process digitally. Furthermore, to stabilize training on limited data samples, we incorporate a linear interpolation strategy into the diffusion process. Extensive experiments on the Food-101 and VireoFood-172 datasets demonstrate that our method achieves state-of-the-art generation quality even in data-scarce scenarios. Crucially, we validate the practical utility of our synthetic images: utilizing them for data augmentation improved the accuracy of downstream food classification tasks from 95.65% to 96.20%. This study provides a cost-effective solution for generating diverse, controllable, and realistic food data to advance smart food systems. Full article

Background: Diagnosing CNS tumors through MRI is limited by significant variability in scanner hardware, acquisition protocols, and intensity characteristics at clinical centers, resulting in substantial domain shifts that lead to diminished reliability for automated models. Methods: We present a Domain-Adaptive MRI Learning Model (DA-MLM) consisting of an adversarially aligned hybrid 3D CNN–transformer encoder with contrastive regularization and covariance-based feature harmonization. Varying sequence MRI inputs ($T1$, $T1ce$, $T2$, and $FLAIR$) were inputted to multi-scale convolutional layers followed by global self-attention to effectively capture localized tumor structure and long-range spatial context, with domain adaptation that harmonizes feature distribution across datasets. Results: On the BraTS 2020 dataset, we found DA-MLM achieved 94.8% accuracy, 93.6% macro-F1, and 96.2% AUC, improving upon previously established benchmarks by 2–4%. DA-MLM also attained Dice score segmentation of 93.1% (WT), 91.4% (TC), and 89.5% (ET), improving upon 2–3.5% for CNN and transformer methods. On the REMBRANDT dataset, DA-MLM achieved 92.3% accuracy with segmentation improvements of 3–7% over existing U-Net and expert annotations. Robustness testing indicated 40–60% less degradation under noise, contrast shift, and motion artifacts, and synthetic shifts in scanner location showed negligible performance impairment (<0.06). Cross-domain evaluation also demonstrated 5–11% less degradation than existing methods. Conclusions: In summary, DA-MLM demonstrates improved accuracy, segmentation fidelity, and robustness to perturbations, as well as strong cross-domain generalization indicating the suitability for deployment in multicenter MRI applications where variation in imaging performance is unavoidable. Full article

Codon optimization is widely used to improve heterologous gene expression in Escherichia coli. However, many existing methods focus primarily on maximizing the codon adaptation index (CAI) and neglect broader aspects of biological context. In this study, we present ColiFormer, a transformer-based codon optimization framework fine-tuned on 3676 high-expression E. coli genes curated from the NCBI database. Built on the CodonTransformer BigBird architecture, ColiFormer employs self-attention mechanisms and a mathematical optimization method (the augmented Lagrangian approach) to balance multiple biological objectives simultaneously, including CAI, GC content, tRNA adaptation index (tAI), RNA stability, and minimization of negative cis-regulatory elements. Based on in silico evaluations on 37,053 native E. coli genes and 80 recombinant protein targets commonly used in industrial studies, ColiFormer demonstrated significant improvements in CAI and tAI values, maintained GC content within biologically optimal ranges, and reduced inhibitory cis-regulatory motifs compared with established codon optimization approaches, while maintaining competitive runtime performance. These results represent computational predictions derived from standard in silico metrics; future experimental work is anticipated to validate these computational predictions in vivo. ColiFormer has been released as an open-source tool alongside the benchmark datasets used in this study. Full article