Search Results (66,734)

Spin–Orbit Torque Magnetic Random-Access Memory (SOT-MRAM)-based in-memory computing (IMC) offers a transformative solution for energy-efficient edge intelligence, yet the deployment of robust online learning remains challenging due to memristive non-idealities and the “memory wall” inherent in Von Neumann architectures. In this work, we propose a Tiki-Taka-inspired SOT-MRAM IMC architecture with a cross-layer co-design that decouples gradient accumulation and weight storage. A time-multiplexing (Time-MUX) strategy is introduced to alleviate interconnect overhead, achieving over 28 times area reduction while maintaining comparable energy consumption to spatial mapping schemes. By integrating a canted-type x SOT-MTJ compact model into the NeuroSim framework, we perform system-level evaluation under realistic constraints. The results reveal that the proposed approach enables an effective trade-off between area, latency, and accuracy. Furthermore, the 1-bit/4-bit hybrid precision scheme achieves accuracy comparable to 8-bit configurations with reduced hardware cost while supporting a five-year cumulative write load of 1.87 × 10⁷ cycles, significantly surpassing the endurance limits of conventional memristor devices. This work provides key insights into cross-layer optimization and establishes a practical design framework for resource-constrained edge learning systems. Full article

In this study, a microcrystalline cellulose (MCC)-stabilized Pickering emulsion approach was developed to integrate hydrophobic natural deep eutectic solvents (NADES; menthol:decanoic acid, 1:1 molar ratio) into agar-based biopolymer films. MCC was evaluated not only as a filler but also as a functional interfacial component governing hydrophobic phase distribution and structural organization. SEM analysis showed that MCC concentration significantly influenced morphology; films with 0.2% MCC exhibited a more homogeneous structure, whereas 0.5% MCC led to heterogeneous and irregular formations. Mechanically, films with 0.2% MCC showed higher elongation at break (16.37%) compared to 0.5% MCC (9.86%), while tensile strength remained similar (2.75–2.78 MPa). Increased MCC content enhanced surface hydrophobicity, as indicated by higher contact angle values. The 0.5% MCC films exhibited high moisture content (85%) and water solubility (93%), attributed to increased free volume and structural irregularity. Swelling index exceeded 40% in 0.2% MCC films but decreased at higher MCC levels. HS-GC-MS analysis revealed temperature-dependent controlled release of menthol, with significant release at 50 °C compared to 25 °C. Antimicrobial tests demonstrated broad-spectrum activity (8.9–24.2 mm). These results highlight MCC as an effective stabilizer for hydrophobic NADES integration and support the potential of these films for active packaging applications. Full article

Intestinal tuberculosis (ITB) is a complex form of extrapulmonary tuberculosis characterized by nonspecific gastrointestinal symptoms and substantial overlap with conditions such as Crohn’s disease and gastrointestinal malignancies. These similarities frequently lead to diagnostic uncertainty, delayed diagnosis, and inappropriate management. This systematic review summarizes current evidence on the clinical presentation, diagnostic approaches, and therapeutic strategies for ITB, with particular emphasis on emerging diagnostic technologies and their role in reducing surgical interventions. A systematic literature search was conducted using PubMed, Scopus, and Google Scholar following PRISMA guidelines to identify relevant studies published from 2000 to 2025. The review focused on clinical manifestations, imaging findings, endoscopic features, histopathological characteristics, molecular diagnostics, pharmacological therapy, and minimally invasive therapeutic interventions. Accurate diagnosis requires an integrated approach combining clinical assessment with imaging, endoscopic evaluation, and histopathological confirmation. Molecular techniques such as GeneXpert MTB/RIF and GeneXpert MTB/RIF Ultra and multiplex polymerase chain reaction assays improve diagnostic accuracy and shorten detection time. Emerging technologies including artificial intelligence-assisted radiologic interpretation and CRISPR-based stool sequencing platforms show promise for earlier detection. Standard anti-tubercular therapy remains the cornerstone of treatment, while minimally invasive endoscopic and surgical procedures are effective for managing complications such as strictures, obstruction, and perforation. Early and precise diagnosis of intestinal tuberculosis is essential to prevent complications and optimize patient outcomes. Integrating conventional diagnostic approaches with emerging molecular and artificial intelligence-based technologies may enhance diagnostic precision and support individualized treatment strategies. Further ITB-specific clinical studies are needed to validate novel diagnostic tools and refine therapeutic approaches for improved patient care. Full article

Obesity and obstructive sleep apnea (OSA) frequently coexist and synergistically contribute to cardiovascular disease through interconnected mechanical, metabolic, and autonomic mechanisms. This interplay promotes myocardial electrical instability and structural remodeling, favoring the development and persistence of cardiac arrhythmias, particularly atrial fibrillation. Among the key mediators linking obesity to arrhythmogenesis, epicardial adipose tissue has emerged as a relevant factor that may contribute to local pro-inflammatory, pro-fibrotic, and autonomic effects on the myocardium. In parallel, OSA-related intermittent hypoxia and intrathoracic pressure swings further amplify electrical instability and autonomic imbalance, reinforcing a self-sustaining arrhythmogenic substrate. Therapeutic strategies are increasingly shifting toward upstream interventions targeting these underlying mechanisms. Metabolic therapies, including the dual GIP/GLP-1 receptor agonist tirzepatide, have demonstrated substantial weight reduction and improvement in OSA severity, with potential indirect benefits on arrhythmic risk through modulation of visceral adiposity, inflammation, and metabolic dysfunction. On the electrophysiological side, cardioneuroablation has emerged as a potentially investigational option in selected patients with vagally mediated bradyarrhythmias, although its role remains to be fully defined. Overall, these observations support an integrated, phenotype-driven approach combining respiratory therapy, metabolic modulation, and targeted electrophysiological interventions. This framework may help redefine therapeutic priorities, shifting from symptom control toward modification of the underlying arrhythmogenic substrate and improvement of long-term cardiovascular outcomes. Full article

Background and Clinical Significance: Early loss of pacing capture after pacemaker generator replacement is an uncommon but potentially life-threatening event, especially in pacemaker-dependent patients. In this setting, device malfunction is often initially attributed to intrinsic lead damage, prompting consideration of invasive lead revision or extraction. However, not all early failures reflect true structural lead dysfunction. Careful interpretation of device interrogation findings, particularly in relation to pacing configuration, may uncover reversible causes and support a more targeted diagnostic and management approach; Case Presentation: A 61-year-old man with complete atrioventricular block presented with recurrent syncope six days after elective pacemaker generator replacement. The electrocardiogram showed absence of effective ventricular pacing with a slow escape rhythm. Device interrogation revealed loss of ventricular capture in bipolar configuration associated with markedly elevated impedance, initially raising concern for lead malfunction. However, switching to unipolar pacing restored effective capture with normal electrical parameters, suggesting preserved lead integrity and prompting reconsideration of the underlying mechanism. Further diagnostic evaluation, including imaging and intraoperative assessment, was therefore undertaken to clarify the cause and guide management; Conclusions: Early pacing failure should not automatically be equated with lead damage. Beyond documenting a reversible lead–header interface problem, this case highlights the diagnostic value of a stepwise approach integrating pacing configuration behavior, targeted imaging, and intraoperative header-independent testing. Such an approach may facilitate rapid localization of reversible defects and help avoid unnecessary lead revision. Full article

The management of deep subgingival carious lesions presents significant challenges for achieving durable adhesive restorations due to limited access, moisture control, and proximity to periodontal tissues. Two main approaches are currently adopted to manage these cases: Deep Margin Elevation (DME) and Surgical Crown Lengthening (SCL). This systematic review (PROSPERO registration CRD420250654262) aimed to compare the performance and survival of restorations placed following DME versus SCL in teeth with deep subgingival margins. A comprehensive literature search was conducted in PubMed, B-ON, and the Cochrane Library for studies published between 2014 and 2025. Following PRISMA guidelines, six studies were included. Methodological quality and risk of bias were assessed using ROBINS-I, RoB 2, and the CARE guidelines. The available evidence indicates that both DME and SCL provide satisfactory periodontal stability, high restoration survival rates, and a low incidence of recurrent caries. DME emerged as a minimally invasive strategy that facilitates adhesive procedures by relocating deep margins to more accessible positions, potentially improving marginal integrity while preserving tooth structure and gingival architecture, particularly in patients with a thick gingival biotype. The choice between DME and SCL should be individualized. Further long-term clinical studies are required to clarify their impact on adhesive interface durability in subgingival environments. Full article

Industrial Control Systems (ICSs) face dual imperatives: protecting critical infrastructure from escalating cybersecurity threats while reducing the environmental impact of AI-powered defense mechanisms. Current deep learning anomaly detection approaches achieve security performance but consumes substantial computational resources, creating an environmental paradox in which AI solutions designed to protect infrastructure contribute to carbon emissions at scale. This competition between cybersecurity effectiveness and sustainability objectives intensifies as regulatory frameworks increasingly mandate both security resilience and environmental accountability. This research presents Green-USAD, a sustainability-first AI framework that inverts traditional design paradigms by integrating energy efficiency as a primary architectural constraint from inception rather than applying compression retrospectively. The proposed approach advances green computing for critical infrastructure through four key contributions: (1) a compressed architecture with validation-guided convergence protocols achieving competitive detection performance with minimal computational overhead; (2) a multi-objective optimization framework using the Analytic Hierarchy Process to systematically balance security and sustainability requirements; (3) a hardware-validated energy measurement methodology addressing reproducibility challenges in green AI literature; and (4) a comprehensive evaluation demonstrating cross-datasets and edge-deployment viability. Validation on ICS benchmarks demonstrates that sustainability-first design achieves substantial energy reduction while maintaining operational detection accuracy, with measured training consumption below 1% of conventional approaches and proportional carbon emission reductions. Comparative analysis against post hoc compression baselines establishes fundamental advantages of design-from-inception over train-then-compress paradigms. Edge device deployment on resource-constrained hardware confirms real-world applicability for distributed industrial environments. Results establish that robust cybersecurity and environmental sustainability represent unified rather than competing objectives when intelligent systems are designed with sustainability as a foundational principle. Full article

The development of supersonic aircraft presents significant challenges in ensuring safety during early design stages, particularly for fuel tank systems exposed to extreme thermal and structural loads. Conventional document-based zonal safety analysis methods are limited in their ability to capture dynamic interactions between spatial subsystem configurations and functional system behavior during early conceptual design, leading to delayed hazard identification. This study proposes an integrated framework combining computer-aided design (CAD) and model-based systems engineering (MBSE) to support early-stage zonal hazard analysis. The framework links spatial subsystem modelling with functional system architecture to enable iterative hazard identification and mitigation. Applied to the SA-24 Phoenix conceptual supersonic aircraft, the approach identifies critical risks, including fuel vaporization, over-pressurization, and structural fatigue, and evaluates mitigation strategies such as thermal insulation and redundant venting. Functional hazard analysis and fault tree analysis are used to assess failure scenarios and ensure compliance with EASA CS-25 requirements. Results indicate an estimated reduction of up to 40% in risk priority number (RPN) values for key thermal hazard pathways and a 25% reduction in conceptual design iteration time compared with conventional approaches. The findings demonstrate that CAD–MBSE integration offers a scalable and efficient methodology for early hazard identification, contributing to safer and more reliable supersonic aircraft design. Full article

Deep learning-based object detectors, particularly You Only Look Once (YOLO) architectures, have demonstrated strong performance in automated brain tumor detection. However, the impact of resolution scaling on tumor localization accuracy remains underexplored, especially under conditions where image resolution is reduced. This study aims to investigate how lowering the input resolution from 640 × 640 to 480 × 480 affects detection performance and whether optimized depth/width scaling and hyperparameter tuning can compensate for the expected loss of spatial detail. In this work, we propose an optimized YOLO-based framework for brain tumor detection and localization in MRI scans, building upon the method “Addressing the Impact of Resolution Scaling on YOLO Performance for Brain Tumor Detection through Optimized Network Depth/Width Adjustments.” Our model, an enhanced variant of the BGF-YOLO architecture, is specifically tailored for the challenges of medical imaging. The proposed network features both architectural and training-level optimizations. We used a publicly available dataset from Kaggle that consists of 500 training images, 201 validation images, and 100 test images. Experimental analysis demonstrates that while reducing input resolution alone degrades performance, integrating targeted modifications specifically increases network depth and width. In addition, advanced training strategies such as MixUp augmentation, dropout regularization, AdamW optimization, cosine learning rate scheduling, and finely tuned learning rate ranges lead to substantial performance gains. The optimized model achieves a precision of up to 0.858, a recall of 0.943, mAP${}_{50}$ of 0.946, and mAP${}_{50-95}$ of 0.672. These results not only outperform the reduced-resolution baseline but also approach, and in some cases surpass, the original high-resolution BGF-YOLO setup. Full article

Ventricular repolarization abnormalities are among the most frequent electrocardiographic findings in pediatric athletes undergoing cardiovascular screening, yet their clinical significance remains a major source of diagnostic uncertainty. While most of them represent benign expressions of training-induced cardiac remodeling and developmental maturation, selected patterns may constitute the earliest phenotypic manifestation of cardiomyopathies or primary electrical disease. Distinguishing physiological adaptation from early pathology is therefore essential to prevent both sudden cardiac events and unnecessary restrictions on sports participation. This review integrates contemporary international electrocardiographic interpretation criteria with emerging pediatric evidence to provide a clinically oriented framework for evaluation and risk stratification of ventricular repolarization abnormalities in pediatric athletes. Early repolarization and anterior T-wave inversion are commonly benign when occurring within recognized age- and ethnicity-specific patterns and in the absence of symptoms, concerning family history, or structural abnormalities. Conversely, lateral or inferolateral T-wave inversion, atypical ST-segment morphology, complex ventricular arrhythmias, and abnormal imaging findings represent red flags requiring comprehensive investigation, including multimodality imaging when indicated. Due to the dynamic electrophysiological evolution during adolescence, longitudinal reassessment is crucial. A structured, risk-based approach integrating electrocardiographic features, demographic/familial context, clinical evaluation, imaging findings, and follow-up provides a pragmatic strategy to optimize risk detection while safeguarding appropriate athletic participation in young athletes. Full article

[Problem] Converting viral tourism popularity into long-term destination sustainability is a central governance challenge in the digital era. [Aim] This study aims to explicitly measure how emotional value mediates the transition from ephemeral online traffic to durable offline place attachment. [Methodology] Adopting a descriptive mixed-methods approach, data were collected through semi-structured interviews with 16 purposively selected participants (including tourists and locals) recruited via on-site intercepts and online snowball sampling. The inclusion criterion required active engagement with Harbin’s digital tourism discourse. Qualitative transcripts were coded using the NVivo 12 software and subsequently converted into panel data. Grey Panel Relational Clustering was then utilized to geometrically track tourist emotional trajectories. [Results] The analysis identified three structural tourist typologies—the Full-Link Empathy Type, Pragmatic Verification Type, and Traffic-Driven Co-conspirator Type—and revealed three corresponding synergistic paths driving online–offline integration: Virtual–Real Isomorphism, Complementarity, and Symbiosis. [Conclusions] The findings demonstrate that sustainable destination resilience depends fundamentally on the qualitative composition of emotional engagement across different tourist types, rather than sheer visitor volume. [Implications] This study contributes an empirically grounded, emotional value-driven framework to sustainable tourism theory, offering differentiated governance strategies for destinations navigating the volatility of platform-driven attention economies. Full article

Model-based reinforcement learning (MBRL) is a promising approach for achieving high sample efficiency in learning control policies. The existing world models in MBRL typically represent the environment’s state as a single global latent vector. However, such representations limit the model’s ability to capture object interactions and reason about individual objects—capabilities that are critical for visual object-oriented tasks—and may lead to lower sample efficiency. To address this limitation, we propose Kuramoto Object-Centric Reinforcement Learning (KORL), a model-based agent that learns an object-centric world model. Our approach introduces a novel Kuramoto Slot Attention for Video (KSAVi) model that integrates Kuramoto oscillatory neurons with the Slot Attention module to robustly extract object representations. We design a world model that leverages these structured object-centric latents and predicts dynamics using graph neural networks, thereby incorporating an inductive bias for modeling object interactions. We evaluate KORL on a suite of visually diverse object-oriented robotic manipulation tasks and demonstrate that our method outperforms object-centric model-free and model-based approaches. Full article

Understanding how medicinal plant distributions shift in response to climate change is essential for developing forward-looking conservation strategies. Cibotium barometz (L.) J. Sm., a tree fern from the family Dicksoniaceae, is not only ecologically significant but also holds considerable medicinal value. Despite its importance, wild populations of this species have been steadily declining due to ongoing habitat loss and unsustainable harvesting. To address this concern, we constructed a multi-model ensemble framework that integrated nine different algorithms, including Generalized Linear Models, various machine learning approaches, and a MaxEnt model optimized through ENMeval using a regularization multiplier of 2 and a feature class of LQH. Using this modeling framework, we simulated the habitat suitability dynamics of C. barometz under current climate conditions (1970–2000) and two future periods (2050s and 2090s) across four Shared Socioeconomic Pathways (SSP126, SSP245, SSP370, and SSP585). Our analysis identified water availability and low temperature stress as the primary factors limiting the species’ distribution. The suitable range for precipitation during the driest quarter extends from 3.25 to 640.20 mm, with optimal conditions occurring when precipitation reaches at least 96.84 mm. Annual precipitation suitable for the species lies between 74.58 and 4209.60 mm, and the most favorable range falls between 3834.10 and 4209.60 mm. While the minimum temperature of the coldest month can vary from −35.41 to 22.35 °C, optimal survival requires temperatures of 8.79 °C or higher. In addition, the species grows best within an annual temperature range of 16.25 to 27.92 °C, with an optimum around 20.47 °C. Projections based on the multi model ensemble suggest that future climate warming may lead to a southwestward shift in the centroid of suitable habitat for this species. By the 2090s, under the SSP245, SSP370, and SSP585 scenarios, the centroid shifts southwestward by 331.3 km, 335.1 km, and 180.2 km, respectively. Meanwhile, areas with high habitat suitability are expected to retreat toward mid-to-high elevation zones, especially in southeastern Yunnan, southern Guizhou, and western Guangxi. The effects of different emission pathways vary considerably; under the high-emission SSP585 scenario, the reduction in total suitable area is projected to be more severe and habitat fragmentation more extensive compared to the low-emission SSP126 pathway. In contrast, implementing ambitious emissions reduction measures could play a key role in supporting the long-term stability of C. barometz populations. This study clarifies how this species responds to climate change and the spatial strategies it may adopt, providing a scientific basis and spatial references for conserving its germplasm resources, restoring its habitats, and advancing its sustainable use. Full article

Background: Hospice refers to specialised end-of-life care that supports patients and families, making it an important area for studying how language shapes experiences and expectations of care. This study compares hospice discourse on websites in Hong Kong and the United Kingdom, analysing how NLP-based sentiment and interpersonal features, such as personal pronouns and conjunctions, shape logical relations, structure information, and express emotion in patient narratives. Methods: Using a mixed approach that integrates sentiment analysis with Systemic Functional Linguistics (SFL), and taxonomy of conjunctions in particular, this study draws on a 52,086-word corpus from 40 hospice websites (20 from each region). The corpus analytical tool AntConc was used to identify co-occurrence, interpret log-likelihood, and perform concordance analysis. Results: The findings reveal significant differences in the digital delivery of hospice care across regions. According to our data, UK websites tend to express a wider range of personal emotions and frequently use concessive conjunctions when discussing sensitive palliative care topics. In contrast, Hong Kong websites tend to use more additive and causal conjunctions, projecting a stronger focus on institutional care. For example, Hong Kong texts tend to use formal, service-oriented connections such as “we + offer”, reflecting a more informational communicative style. However, both regions frequently use personal pronouns such as “you” and “we” to convey positive sentiment and demonstrate empathy towards patients and their caregivers. Conclusion: These patterns appear to be used strategically by hospice providers to build trust, signal alignment, and strengthen relationships tailored to each region. Lastly, this study makes an original contribution by combining computational and functional linguistic approaches to develop a systematic method for examining culturally shaped digital communication in end-of-life contexts, thereby enriching the field of healthcare discourse analysis. Full article