Search Results (270)

Remote sensing semantic segmentation is fundamental for fine-grained urban scene understanding, which in turn provides pixel-level semantic insights for urban development and environmental surveillance. However, existing hybrid segmentation architectures fail to incorporate intrinsic geometric and physical priors, inevitably leading to structural fragmentation, boundary ambiguity, and spatial misalignment of heterogeneous features. Therefore, we propose a Superpixel-Tokenized and Frequency-Modulated Hybrid CNN–Transformer network (SFCT-Net) for remote sensing semantic segmentation. The proposed network integrates superpixel tokens and high-frequency constraints to preserve structural integrity and boundary precision. First, our Superpixel-Tokenized Linear Position Attention (STLPA) module replaces rigid window tokens with semantic superpixels to ensure object integrity with linear computational complexity. Second, we construct a Frequency-Modulated Deformable Edge Refinement (FMDER) module that leverages high-frequency spectral priors to modulate deformable sampling, achieving robust boundary recovery. Finally, we develop the Spatial–Semantic Feature Coupling (SSFC) module, which employs a dual-branch strategy to correct spatial drift and align deep semantic features with shallow details. Experiments conducted on our self-built Taiyuan Satellite Remote Sensing Dataset (TSRSD) along with the ISPRS Vaihingen and Potsdam benchmark datasets demonstrate that our proposed SFCT-Net delivers state-of-the-art performance and efficiency by fusing superpixel and frequency priors for robust structural and boundary recovery. Full article

Time series data find extensive applications in finance, healthcare, and industrial monitoring domains. However, analytical models targeting such data are subject to notable constraints imposed by the rigid independent and identically distributed (IID) assumption and the high cost of data annotation. Unsupervised Domain Adaptation (UDA) offers an effective remedy for these challenges, and Contrastive Learning (CL) has been widely integrated into UDA frameworks, owing to its robust feature representation and clustering capabilities. Nonetheless, existing CL-based UDA methods suffer from two key limitations: (1) fixed data augmentation strategies result in imbalanced intensity—excessive augmentation erodes sample semantics, while insufficient augmentation induces model overfitting; (2) distribution alignment strategies neglect hard samples which are the core carriers of domain shift, causing their domain adaptation signals to be overshadowed by a large number of normal samples and thus degrading alignment accuracy. To address these drawbacks, this paper proposes a time-series UDA algorithm, termed Adaptive Contrastive Learning Domain Adaptation (ACLDA), which incorporates two key components: (1) an adaptive feature enhancement module that integrates adaptive sample augmentation and CL, enabling the model to capture high-quality transferable features; (2) sample-level adaptive weights, introduced on the basis of class-level alignment via supervised CL, to emphasize the value of hard samples. Comparative experiments on multiple time-series datasets demonstrate that our ACLDA outperforms state-of-the-art domain adaptation methods in terms of average accuracy, verifying its superiority and providing a more robust solution for cross-domain time series analysis. Full article

Wearable health monitoring systems (WHMS) are recognized as key enablers of continuous real-time physiological sensing in healthcare, eldercare, sports, and occupational safety. However, current devices face critical limitations due to their dependence on non-renewable batteries, rigid substrates, and non-degradable electronic components, which contribute to environmental waste and limit long-term usability. This study aims to explore the development of sustainable, energy-autonomous WHMS that integrate multimodal energy harvesting, including triboelectric, piezoelectric, photovoltaic, thermoelectric, and radio frequency, with biodegradable and bioresorbable electronics using silk fibroin, cellulose nanofibers, poly(lactic-co-glycolic acid), magnesium, and transient silicon. This unified system architecture would further comprise harvesters, power management circuits, energy buffers, low-power sensing front-ends, and tiny machine learning-enabled data processing. The methodology emphasizes energy-neutral operation through duty-cycling, harvest-aware scheduling, and compressive sensing. Simulation and modeling results indicate harvested power densities between 100 and 220 µW·cm⁻², sufficient to sustain electrocardiography, photoplethysmography, and temperature monitoring under realistic daily use profiles. Material degradation studies demonstrate predictable dissolution kinetics over 8–20 weeks in physiological conditions, aligning with safety and environmental goals. By uniting sustainable materials science with energy-efficient circuit design, this work establishes a blueprint for the next generation of eco-friendly, clinically relevant, and ethically responsible wearable health technologies. Full article

Recent GraphRAG methods based on knowledge graphs (KGs) primarily rely on either under-reasoning or a structural path-level retriever, which prevents them from jointly capturing fine-grained semantic relevance and explicit multi-hop reasoning paths. This separation often results in semantic mismatch—where logical links are missing—or structural over-constraint in reasoning— where rigid dependencies limit flexible reasoning—thereby degrading both answer accuracy and the reliability of evidence in complex KGQA tasks. To address these issues, we propose HybRAG, a hybrid retrieval framework that synergistically integrates a semantic node-level retriever and structural path-level retriever. HybRAG constructs a hybrid subgraph that jointly reflects the semantic proximity of entities and the relational structures encoded in the KG. Furthermore, we incorporate retrieval-augmented fine-tuning, which enables the model to internalize advanced reasoning strategies for interpreting disparate semantic and structural signals, rather than merely memorizing domain facts. Through extensive experiments on the WebQSP and CWQ benchmarks, we demonstrate that HybRAG effectively bridges the gap between LLM-centric semantic approaches and GNN-centric structural approaches, outperforming single-retriever baselines. Our findings, including detailed sensitivity and ablation analyses, provide empirical evidence that the systematic alignment of semantic and structural signals is essential for ensuring the reasoning reliability and scalability of next-generation GraphRAG systems. Full article

Semantic interoperability remains a critical challenge in Spatial Data Infrastructures (SDIs), particularly when aligning authoritative taxonomies with collaborative folksonomies. Traditional alignment tools often fail to bridge the semantic and structural asymmetry between these schemas. This paper evaluates the capability of Large Language Models (LLMs), specifically distinguishing between traditional architectures and emerging Large Reasoning Models (LRMs), to perform semantic alignment between the Brazilian national topographic data model standard (EDGV) and OpenStreetMap (OSM). Using a formal ontology as a prompting scaffold, we tested seven model versions (including ChatGPT 5, DeepSeek R1, and Gemini 2.5) on their ability to bridge the gap between rigid hierarchical classes and the dynamic, ‘long-tail’ vocabulary of the folksonomy. Results reveal a distinct trade-off: while traditional LLMs exhibited ‘lexical rigidity’ and popularity bias—failing to map low-frequency tags—Reasoning Models demonstrated significantly improved capacity for semantic expansion, correctly identifying complex many-to-one (n:1) relationships across linguistic barriers. However, this reasoning depth often came at the cost of ‘hallucination by over-specification’ and syntactic instability in generating OWL code. We conclude that a neuro-symbolic approach, positioning LRMs as ‘Semantic Catalysts’ within a Human-in-the-Loop (HITL) workflow, provides a viable pathway for interoperability, balancing generative power with the need for logical rigor and spatial validation. Full article

Muscle degeneration is the hallmark of muscular dystrophies—genetically heterogeneous disorders traditionally approached through the lens of molecular pathogenesis or symptomatic management in isolation. Here, we present a deliberately interdisciplinary synthesis that bridges molecular genetics, clinical phenotyping, and evidence-based orthopedic decision-making to address a significant critical gap: the lack of genotype-informed, function-oriented frameworks for musculoskeletal complications. We re-evaluate disease entities—not only by their molecular etiology (e.g., DMD, LMNA, DUX4 dysregulation), but through the prism of orthopedic manifestations as diagnostic gateways and therapeutic milestones. For instance, early rigid spine in LMNA-related dystrophy is not merely a sign of contracture, but a red flag demanding cardiac risk stratification before surgical planning, in alignment with current consensus. Similarly, scoliosis management in Duchenne muscular dystrophy is discussed through quantitative decision thresholds (Cobb angle ≥ 20–30°, FVC ≥ 30–35%) derived from long-term outcome studies, rather than general clinical recommendations. Critically, we confront challenges posed by disease-modifying therapies: patients now survive into their 30s and 40s, yet develop novel, therapy-exacerbated orthopedic phenotypes (e.g., steroid-induced osteoporosis, atypical spinal rigidity). Therefore, we argue that precision orthopedics—tailored surveillance, genotype-stratified intervention timing (e.g., D4Z4 repeat-guided monitoring in FSHD, and realistic functional goal-setting (e.g., scapular arthrodesis for overhead function)—should become the gold standard of care. For example, desminopathies may show marked phenotypic variability even within the same mutation. Our review thus serves not only as a molecular overview, but as a practical roadmap for neurologists, geneticists, orthopedic surgeons, and rehabilitation specialists seeking to translate genomic insights into durable functional outcomes. Full article

Accurate identification of dynamic parameters, specifically natural frequency and damping ratio, is critical for optimizing the disturbance rejection performance of laser level self-leveling mechanisms. However, traditional Finite Element Analysis (FEA) often struggles to quantify micro-friction damping, while contact measurement methods introduce added mass interference. To address these challenges, this paper proposes an integrated framework combining Pulse-Window Software Lock-in (PWSL) sensing with a data-driven model updating strategy. Initially, a rigid-body dynamic model theoretically predicted a natural frequency (f_sim) of 2.987 Hz and a damping ratio (ζ_sim) of 0.1255. To acquire authentic responses, a non-contact Position Sensitive Detector (PSD) system was developed. The custom PWSL algorithm leverages the laser’s 10 kHz carrier to extract high-fidelity displacement signals, effectively suppressing broadband noise despite embedded hardware limitations. Experimental results demonstrated that the measured frequency (f_exp = 2.861 Hz) aligned well with predictions (4.22% error). In contrast, the measured damping ratio (ζ_exp = 0.1435) exceeded the simulation value by 14.34%, quantitatively revealing the energy dissipation caused by unmodeled bearing friction. Based on this disparity, the FEA model was inversely updated by introducing an equivalent friction coefficient, successfully reducing the damping prediction error to 0.97%. This study establishes a high-fidelity updated model, providing a reliable basis for the refined design of precision pendulum instruments. Full article

This study systematically reviews 53 peer-reviewed articles on public sector innovation published between 2021 and 2025 to synthesize knowledge on how innovation is conceptualized, triggered, enacted, and constrained. Findings reveal that innovation is framed across technological, organizational, governance, and social dimensions, reflecting substantial conceptual and theoretical diversity. Key triggers include digital transformation, leadership, inter-organizational collaboration, fiscal pressures, and workforce capabilities, with emphasis shifting toward technology, human capital, and collaboration in recent years. Innovation produces both positive outcomes, such as improved service quality, efficiency, and citizen engagement, and negative or unintended consequences, including implementation failures, equity concerns, and employee resistance. Persistent barriers, such as bureaucratic rigidity, risk-averse culture, accountability pressures, and political interference, operate as structural conditions rather than isolated obstacles. Theoretical foundations remain fragmented, with New Public Management, New Public Governance, institutional theory, and public value theory applied inconsistently. These findings underscore the need for integrative, context-sensitive approaches that combine institutional, human, and technological perspectives to guide innovation effectively. The review offers actionable insights for public managers and policymakers, emphasizing alignment with organizational capacity, leadership, and regulatory design, and highlights directions for future research to advance theory, practice, and policy in public sector innovation. Full article

Background/Objectives: Adolescent idiopathic scoliosis (AIS) is a three-dimensional spinal deformity commonly managed with conservative strategies, including bracing and physiotherapeutic scoliosis-specific exercises (PSSEs). The Canali Postural Method^® (CPM) is an individualized kinesiological approach aimed at postural reprogramming, while the Canali Orthopedic Brace is an intermittent, non-rigid device intended to facilitate active postural control rather than continuous passive correction. Case Presentation: We retrospectively report two adolescent females with thoracolumbar rotoscoliosis (Risser grade 4). Case 1 (15 years) presented with a left-convex thoracolumbar curve (apex T12–L1) with a Cobb angle of 19.4° and a derotation angle ratio (DAR) of 1.9. Case 2 (16 years) presented with a right-convex thoracolumbar curve (apex T10) with a Cobb angle of 41.14° and a DAR of 3.7. Both patients underwent supervised CPM-based exercise sessions combined with intermittent use of the Canali Orthopedic Brace. Discussion and Conclusions: Follow-up radiographs showed a marked reduction in curve magnitude and rotational parameters: in Case 1, the Cobb angle decreased from 19.4° to 4.1° and DAR from 1.9 to 0.4; in Case 2, the Cobb angle decreased from 41.14° to 15.17° and DAR from 3.7 to 1.36. Pelvic asymmetry was also reduced, and no worsening of sagittal alignment was observed. Given the retrospective design, the small sample size, heterogeneity in intervention duration, and the lack of clinical outcomes and formal measurement reliability testing, these findings should be interpreted with caution and warrant confirmation in prospective controlled studies. Full article

Surface defects in hydraulic concrete structures exhibit extreme topological heterogeneity. and are frequently obscured by unstructured environmental noise. Conventional detection models, constrained by fixed-grid convolutions, often fail to effectively capture these irregular geometries or suppress background artifacts. To address these challenges, this study proposes the Adaptive Local–Global Synergistic Perception Network (ALGSP-Net). First, to overcome geometric constraints, the Defect-aware Receptive Field Aggregation and Adaptive Dynamic Receptive Field modules are introduced. Instead of rigid sampling, this design adaptively modulates the receptive field to align with defect morphologies, ensuring the precise encapsulation of slender cracks and interlaced spalling. Second, a dual-stream gating fusion strategy is employed to mitigate semantic ambiguity. This mechanism leverages global context to calibrate local feature responses, effectively filtering background interference while enhancing cross-scale alignment. Experimental results on the self-constructed SDD-HCS dataset demonstrate that the method achieves an average Precision of 77.46% and an mAP50 of 72.78% across six defect categories. Comparative analysis confirms that ALGSP-Net outperforms state-of-the-art benchmarks in both accuracy and robustness, providing a reliable solution for the intelligent maintenance of hydraulic infrastructure. Full article

Industrialised construction (IC) represents a foundational strategy for overcoming entrenched productivity constraints and supply shortfalls in the housing sector. By enabling the mass production and mass customisation of advanced kit-of-parts systems, IC supports more efficient, predictable, scalable, and sustainable building delivery through integrated, standardised, and digitally enabled processes. However, adoption remains uneven due to fragmentation across regulatory, organisational, and technological systems. This paper presents a systematic literature review and thematic synthesis of the literature published between 2000 and 2025 to examine performance outcomes, adoption trends, digital integration maturity, and emerging platform-based design for manufacture and assembly (P-DfMA) approaches, and the main drivers. The review shows that significant performance gains are achievable, including notable reductions in construction time and cost variability, along with substantial reductions in material waste, together with measurable improvements in quality, safety, and delivery predictability. However, widespread uptake of IC remains constrained. This is largely driven by regulatory misalignment, rigid and bespoke procurement and delivery models, inconsistent and unstable supply chain capacity, and the lack of standardised components and integrated digital workflows. Building on these insights, this paper examines the key enablers required for sector-wide transformation toward an ecosystem that supports standardised kit-of-parts solutions, digitally driven design-to-production workflows, and aligned policy and procurement frameworks that are capable of delivering scalable and repeatable industrialised housing. The findings provide a consolidated evidence base and identify the key enablers for policymakers, industry stakeholders, and researchers working to move from project-centred delivery models to platform-based, digitally integrated, and industrialised construction systems. We searched Scopus, Web of Science, ScienceDirect, and Google Scholar, complemented by targeted industry and policy repositories; the searches were last updated on 1 December 2025. After screening, 117 sources were included. The review was not registered, and no review protocol was prepared. Full article

Lactulose, a valuable functional disaccharide with pharmaceutical and food applications, is efficiently synthesized via enzymatic isomerization of lactose. This study developed an integrated strategy combining protein engineering of cellobiose 2-epimerase (CsCE) from Caldicellulosiruptor saccharolyticus and process optimization to enhance lactulose production. A dual-track engineering approach—incorporating flexible loop modulation (residues 161–193) and structure-guided sequence alignment with N-acetyl-D-glucosamine-2-epimerase—enabled the creation of two superior mutants, R17Q/L184S and R17Q/S142T. The R17Q/L184S variant exhibited a 37% increase in crude enzyme activity, improved thermostability (half-life of 200 min at 80 °C), and enhanced substrate affinity (K_m reduced by 23.2%). R17Q/S142T achieved a 21% higher specific activity (24.08 U/mg), the highest among all variants. Structural and molecular dynamics analyses revealed that L184S enriched hydrogen bonding and hydrophobic interactions, improving structural rigidity, while S142T introduced allosteric regulation that facilitated catalytic efficiency. Under optimized conditions (70 °C, pH 7.5, 40% lactose, 20 U/mL enzyme, 3 h), lactulose yield reached 75.6% with >95% purity. This work demonstrates the successful application of synergistic enzyme engineering and process intensification for high-efficiency lactulose biosynthesis, providing viable candidates and system solutions for industrial-scale production. Full article

Education systems worldwide face a growing pressure to align with Sustainable Development Goal 4.7 by embedding Education for Sustainable Development (ESD) into school life. This study examines how primary school headteachers in Cyprus interpret and enact sustainable leadership to advance ESD within a small, highly centralised system. Drawing on sustainable and distributed leadership theories and a whole-school lens, the study employed semi-structured interviews with ten headteachers from diverse regions (urban, rural, and semi-rural). Reflective thematic analysis identified four patterns: (1) leaders sought a strategic integration of ESD into planning and culture; (2) empowerment and participation were pursued through teacher working groups, student eco-councils, and community partnerships; (3) systemic constraints, a rigid curriculum, limited autonomy, and scarce professional development produced a policy–practice gap; and (4) leaders relied on adaptive, collaborative micro-practices to sustain momentum. The findings suggest that, in Cyprus, sustainable leadership operates as a values-based stewardship enacted through ‘quiet activism’. The study highlights implications for leadership development, such as reflexivity, systems thinking, and ethical reasoning, as well as policy design, such as time, autonomy, and structured support for whole-school ESD, in small-state contexts. Full article

Previous studies, constrained by the overly rigid stability requirements, often fail to adapt to complex systems and struggle to identify stable outcomes that align with the practical context of multi-agent resource allocation. To address the three-sided matching problem in complex socio-technical and business management systems, this paper proposes a fuzzy stable matching method for three-sided agents under a framework of combinatorial preference relations, integrating network and decision theory. First, we construct a membership function to measure the degree of preference satisfaction between elements of different agents, and then define the concept of fuzzy stability. By incorporating preference satisfaction, we introduce the notion of fuzzy blocking strength and derive the generation conditions for blocking triples and fuzzy stability under the fuzzy stable criterion. Furthermore, we abstract the three-sided matching problem with combined preference relations into a shortest path problem. Second, we prove the equivalence between the shortest path solution and the stable matching outcome. We adopt Dijkstra’s algorithm for problem-solving and derive the time complexity of the algorithm under the pruning strategy. Finally, we apply the proposed model and algorithm to a case study of project assignment in software companies, thereby verifying the feasibility and effectiveness of this three-sided matching method. Compared with existing approaches, the fuzzy stable matching method developed in this study demonstrates distinct advantages in handling preference uncertainty and system complexity. It provides a more universal theoretical tool and computational approach for solving flexible resource allocation problems prevalent in real-world scenarios. Full article

As new technologies, such as electronic medical records (EMRs), are introduced into healthcare services, we need to consider how they may be incorporated into simulated environments, so as to maintain and enhance authenticity and learning opportunities. While EMRs have revolutionised clinical practice, many education settings continue to rely on paper-based documentation in simulation, creating a widening gap between educational environments and real-world clinical workflows. This disconnect limits learners’ ability to engage authentically with the tools and resources that underpin contemporary healthcare, impeding the transfer of knowledge to the clinical environment. This practice-based commentary draws on institutional experience from a large, multi-disciplinary simulation-based education facility that explored approaches to integrating EMRs into simulation-based education. It describes the decision points and efforts made to integrate an EMR into simulation-based education and concludes that while genuine EMR systems increase fidelity, their technical rigidity and data governance constraints reduce authenticity. To overcome this, Adelaide Health Simulation adopted an academic EMR (AEMR), a purpose-built digital platform designed for education. The AEMR maintains the functional realism of clinical systems while offering the pedagogical flexibility required to control data, timelines, and learner interactions. Drawing on this experience, this commentary highlights how authenticity in simulation-based education is best achieved not through technological replication alone, but through deliberate use of technologies that align with clinical realities while supporting flexible, learner-centred design. Purpose-built AEMRs exemplify how digital tools can enhance both fidelity and authenticity, fostering higher-order thinking, clinical reasoning, and digital fluency essential for safe and effective contemporary healthcare practice. Here, we argue that advancing simulation-based education in parallel with health service innovations is required if we want to adequately prepare learners for contemporary clinical practice. Full article