Search Results (7,511)

Background: Conventional Doppler ultrasound is limited by its angle dependence and its inability to characterize complex flow patterns adequately. Vector flow imaging (VFI) is an angle-independent ultrasound technique that enables two-dimensional velocity vector mapping and the assessment of hemodynamic parameters. This scoping review aims to map the current evidence on the application and validation of VFI for assessing carotid artery stenosis (CAS). Methods: Using the Population–Concept–Context framework, we included studies that evaluated the application, validity, or outcomes of vector flow imaging in individuals with suspected or confirmed carotid artery stenosis, as well as in healthy controls used for comparison or validation, in clinical or research settings. Five electronic databases were searched from inception through 30 April 2026. Studies limited to conventional Doppler ultrasound or to animal or phantom models without human validation were excluded. Data were synthesized using descriptive numerical analysis and narrative thematic synthesis. Results: A total of 48 sources of evidence met the inclusion criteria. Six trial registrations without reported results were excluded from synthesis, leaving 42 synthesized sources, including 40 original studies and two reviews. The most frequently reported parameters were wall shear stress (n = 25), turbulence indices (n = 11), and velocity vectors (n = 14). Most studies were cross-sectional (n = 20) or technical validation studies (n = 13) with small sample sizes (median 31 participants, IQR 10.5–57; range 1–476). The dominant research focus was plaque vulnerability assessment, while fewer studies addressed stenosis grading. Only eight studies reported diagnostic-performance metrics; none employed prespecified thresholds, two reported blinding, and none adhered to the Standards for Reporting Diagnostic Accuracy Studies guidelines (STARD). Conclusions: VFI enables detailed characterization of carotid hemodynamics, but the available evidence remains preliminary and does not establish clinical utility. Standardized, adequately powered studies with external validation are required before clinical implementation. Full article

Understanding intra-urban spatial interaction is essential for context-sensitive urban planning. While classical spatial interaction theories provide strong conceptual foundations, systematically translating these theoretical concepts into quantifiable indicators remains a significant methodological challenge. Furthermore, capturing the complex, non-linear dynamics driving urban mobility requires analytical approaches that balance predictive power with interpretability. To address this gap, this study develops a feasible, theory-informed analytical framework that bridges classical spatial interaction theory with interpretable machine learning to quantify the predictive patterns underlying intra-urban mobility. In a case study of Changchun, China, Ullman’s three core concepts, together with fundamental measures of urban scale, were systematically operationalized as a set of quantitative proxy variables based on multi-source geospatial big data. XGBoost was then used to model grid-level origin-destination flows at multiple spatial resolutions, and the SHapley Additive exPlanations (SHAP) was used to assess the contributions and dependence patterns of the theory-driven indicators. The results demonstrate the framework’s predictive robustness, with the XGBoost model consistently outperforms the traditional parametric benchmark across all evaluated spatial resolutions. The 1000 m resolution provided the best balance between predictive performance and spatial detail, yielding an R² of 0.696, compared with 0.611 for the benchmark. The explanatory analysis indicates that functional complementarity is the most critical predictive dimension overall. It also identifies distinct nonlinear patterns, including negative associations between transfer impedance and predicted mobility flows beyond critical thresholds, positive associations between built-environment scale indicators and predicted flows only above minimum intensity thresholds, and diminishing marginal associations between intervening opportunities and predicted flows. This study provides a scalable and transferable approach for diagnosing spatial interactions in data-rich urban contexts, providing an empirical basis for calibrating future micro-level urban simulations. Full article

Cooling-water intake basins in coastal power plants may experience particle retention and sediment deposition when complex inlet geometry produces large-scale recirculation and low-velocity zones near intake structures. This study examines the use of submerged barriers to mitigate particle retention in a de-identified coastal power-plant intake basin using hydraulic model experiments and hydraulic-model-supported three-dimensional computational fluid dynamics (CFD). A geometrically consistent model-scale configuration, including the inlet channel, main basin, three intake openings, and two outlet passages, was used to preserve site confidentiality while retaining the essential hydraulic mechanisms. Surface-flow patterns, water-depth variations, and sediment-deposition behavior were measured in the physical model and used to assess the numerical model. The numerical results, supported by the available hydraulic-model observations, reproduced the dominant counterclockwise recirculation and a broadly similar retention-prone region. The simulated water depths agreed closely with the measurements, with relative errors below 0.50% for the finest mesh. A water-depth-based grid-sensitivity assessment using 1,221,165; 2,414,216; and 3,378,840 computational cells further showed that the predicted mean water depths remained within 1.15% of the experimental measurements. The assessed numerical model was then applied to examine submerged-barrier configurations installed near the inlet-to-basin transition, with the barrier-performance interpretation limited to the tested mesh, the assumed representative particle condition, and the available qualitative flow/deposition evidence. Under this assumed particle-tracking condition, the original configuration retained 6989 particles at t = 200 s, whereas the 6 cm submerged barrier reduced the retained-particle count to 4956, corresponding to a reduction of approximately 29.1%. In contrast, the 15 cm emergent barrier increased the retained-particle count to 7573 because it blocked overtopping flow and induced new separated low-velocity regions. These results indicate that, among the tested configurations and under the assumed representative particle condition, the 6 cm submerged barrier yielded the lowest retained-particle count, whereas an excessively high barrier may deteriorate the internal flow structure and increase particle accumulation. Full article

Background: Cerebral blood flow (CBF) is essential for maintaining cerebral metabolism in extremely preterm infants; however, no universally accepted reference standard exists for CBF velocity (CBFV). This study aimed to establish physiological reference patterns of middle cerebral artery CBFV under strictly defined physiological stability using multimodal monitoring. Methods: This post-hoc analysis included extremely preterm infants (gestational age ≤ 28⁺⁶ weeks or birthweight 500–1000 g) from a prospective cohort study. Serial Doppler ultrasonography of the middle cerebral artery was performed along with near-infrared spectroscopy and electrical cardiometry. Normal CBFV datasets were defined based on systemic stability, adequate cardiac output (>150 mL/kg/min), normal regional cerebral oxygen saturation (65–85%), and the absence of major comorbidities. Associations between Doppler parameters and postmenstrual age (PMA) and concurrent weight were analyzed using generalized estimating equations. Results: Forty infants contributed 194 normal CBFV datasets. Physiological reference ranges for peak systolic velocity (PSV), end-diastolic velocity (EDV), mean velocity (MV), resistance index (RI), and pulsatility index (PI) were established across PMA and concurrent weight strata. Both PSV and MV were associated with PMA and concurrent weight (p < 0.01), whereas EDV, RI, and PI were not. Mean blood pressure was not significantly associated with CBFV parameters. Conclusions: PSV and MV were maturation-dependent cerebral perfusion markers. Multimodally defined physiological stability provides a robust framework for establishing clinically relevant reference values for middle cerebral artery CBFV in extremely preterm infants. Full article

Hybrid CNN–Transformer (HCT) synthesis for automated neurodevelopmental diagnostics is an effective approach to constructing intelligent detection systems that are not merely oriented toward feature classification but primarily toward solving spatiotemporal pattern recognition problems in motor disorder assessment. In neurodevelopmental diagnostics, existing automated methods rely on fixed, single-model architectures that process spatial or temporal motion features independently, failing to adapt to the heterogeneous motor irregularities characteristic of developmental coordination disorder and degrading detection sensitivity and generalization across diverse patient populations. There is therefore a pressing need for models capable of simultaneously capturing intra-frame spatial coordination patterns and inter-frame temporal movement dependencies against interrelated diagnostic criteria including accuracy, sensitivity, and motor irregularity specificity. To address this challenge, this paper proposes HCT, a novel framework that integrates ResNet-based spatial feature extraction from optical flow maps and pose estimation skeletons with multi-head self-attention Transformer encoding for modeling long-range temporal dependencies across multi-frame motion sequences. Unlike conventional single-stream approaches, where spatial and temporal processing remain confined to independent architectures, HCT decouples spatiotemporal feature learning through a cross-modal fusion pipeline, constructing a unified discriminative architecture that captures motor coordination dependencies between motion imaging inputs and multiple diagnostic criteria simultaneously. The convolutional encoder generates diverse joint displacement features, which are consolidated through cross-modal attention fusion into a robust, unified embedding with enhanced generalization and resilience to inter-individual motor variability. Integration within neurodevelopmental assessment frameworks facilitates reliable developmental coordination disorder classification, motor irregularity prediction, and interpretable diagnostic decision support, advancing the accuracy, flexibility, and clinical validity of intelligent motor disorder diagnostic systems. Full article

Accurate identification of multiphase flow patterns in vertical pipelines is critical for operational safety and efficiency in the oil and gas industry. Yet, conventional methods struggle with subjectivity and transitional regimes. This study develops and integrates three neural network architectures: a convolutional neural network (CNN) for spatial features, a transformer neural network (TNN) for long-range dependencies, and a physics-encoding network (PEN) for embedding physical constraints. These are combined into a hybrid framework trained on an experimental dataset of 2131 images from a wire mesh sensor, annotated using a semi-automated pipeline. Results show the hybrid model achieved 95.91% test accuracy with a macro F1-score of 0.96, the highest of the four models evaluated, with its main advantage in transitional regimes. A multi-seed ablation shows that the convolutional branch provides the dominant discriminative signal, while the transformer and physics-inspired branches added complementary improvements that are consistent across runs. This hybridisation mitigates individual model weaknesses, with the physics-inspired branch acting as a spatial regulariser that improves interpretability, providing a robust and objective tool for reliable pipeline monitoring. Full article

The cellular immune response underlying post-injury multiple organ dysfunction syndrome (MODS) remains incompletely described. We hypothesized that early perturbations in innate lymphocyte behavior are critical to the development of MODS in trauma patients. To address this, we examined lymphocyte subsets in a prospective cohort of major trauma patients recruited at a single major trauma hospital. Circulating lymphocytes were profiled with flow cytometry in serial samples drawn within the hyperacute (≤2 h) and acute (24 h, 72 h) post-injury periods. Plasma levels of specific mediators derived from innate lymphocytes were also measured in a larger cohort. We observed a marked hyperacute increase in circulating NK (particularly the CD56^dim subset) and Vδ1 cells that were associated with MODS or early mortality. Absolute counts of NK activation markers CD69 and NKG2D were also higher in patients with adverse outcomes, although the proportion of NK cells expressing NKG2D was reduced. Exploratory cluster analyses of NK activating and inhibiting receptors identified patient groups with differing injury characteristics and outcomes. In plasma, patients who developed MODS had significantly higher levels of NK-associated cytotoxic mediators and cytokines. Collectively, these data indicate a specific pattern of hyperacute NK cell activation after major trauma that is characterized by a pattern consistent with cytotoxic lymphocyte activation and is associated with clinical outcome. Full article

Roof-contact backfilling is a critical determinant of stope stability in cut-and-fill mining, and the rheological properties of backfill slurry decisively influence the quality of roof contact. To investigate the flow characteristics of unclassified tailings backfill slurry and their effect on rheological parameters, this study uses the Daye Iron Mine as its engineering case. It adopts a combined laboratory and numerical simulation approach. The physicochemical characteristics of the unclassified tailings and the rheological behavior of the slurry were systematically characterized using particle-size analysis, density measurements, spreadability tests, and rheometer measurements. Subsequently, a numerical model of the L-type flow tester was developed in COMSOL Multiphysics (6.4) to simulate the flow process at varying concentrations. Based on the simulation results, a Gaussian process regression (GPR)-based inversion model for rheological parameters was proposed, and the predictive performance of different kernel functions was compared and evaluated. Finally, the existing backfilling scheme at the Daye Iron Mine was optimized based on the obtained rheological characteristics to improve the roof-contact rate. The results indicate that the unclassified tailings from the Daye Iron Mine have a median particle size of 12.1 μm and a density of 2855 kg·m⁻³, with CaO, Al₂O₃, and MgO as the primary active components. Under the same cement-to-tailings ratio, slurry flowability decreases markedly with increasing concentration. The rheological curves exhibit three stages, with the third conforming to the Bingham model; both yield stress and viscosity increase exponentially with concentration. Evaluation of the inversion results demonstrates that the GPR model with the Rational Quadratic (RQ) kernel achieves optimal performance. The recommended slurry concentration for the Daye Iron Mine is determined to be in the range of 69–71%, and the recommended spacing between filling pipelines is 13.34–18 m. This study reveals the flow evolution patterns of unclassified tailings backfill slurry, demonstrates the potential of the GPR-based inversion approach, and optimizes the roof-contact backfilling scheme, offering a scientific reference for flow characterization and backfill optimization in analogous mining operations. Full article

The urban industrial spatial structure is widely recognized as a critical factor shaping public transit travel patterns. However, the relationship between variations in this spatial structure and bus origin-destination (OD) passenger flows remains insufficiently explored. To address this gap, this study proposes two quantitative indicators, namely, Regional Industrial Proximity (RIP) and Regional Industrial Scarcity (RIS), to characterize the differentiation of urban industrial spatial structures. The empirical study is conducted in the core built-up area within Beijing’s 5th Ring Road. The study area is divided into 8.3 km × 8.3 km grids, and a Proximity–Scarcity Relational Graph Convolutional Network (PS-RGCN) model is employed to predict urban bus passenger flows. The dataset primarily comprises approximately 4.21 million smart card transaction records collected over one continuous week, alongside 276,000 Points of Interest (POI) records covering 20 industrial categories. Specifically, the proposed PS-RGCN model achieves the best prediction performance among all benchmark models. The best overall performance is attained when incorporating the RIP of Scenic Spots and Historical Sites as edge relationships, yielding a coefficient of determination R² of 0.888. In comparison, the gravity model yields an R² of 0.858, and the baseline GCN model yields an R² of 0.378, indicating the superior predictive capability of the proposed model. This study verifies that industrial proximity and industrial scarcity exert complementary effects in bus OD flow prediction. Incorporating both factors synergistically into the multi-relational graph neural network framework enables more effective identification of the predictive relationship between urban functions and bus travel demand. Full article

To address the core industrial bottlenecks of conventional batch tannic acid-based germanium precipitation processes—high reagent consumption, long reaction cycles of several hours, severe impurity co-precipitation as well as the common mismatch between single-channel microreactor throughput and industrial production demands. This work combines numerical simulation with experimental validation to investigate microscale two-phase flow regulation, high-throughput microreactor optimization, and tannic acid precipitation intensification. Two-dimensional two-phase flow models are established for straight and zigzag microchannels, with the level set method applied to track interfacial evolution. The regulatory effects of inlet velocity and channel geometry on flow patterns, droplet behavior and mixing performance are clarified. Zigzag channels induce chaotic convection via periodic corners, achieving an order-of-magnitude improvement in mixing efficiency at low Reynolds numbers (Re < 400), which lays a fundamental basis for reaction intensification. Taking zigzag channels as core units, a bidirectional symmetric superposition scale-up strategy is proposed to break the throughput limitation of single-channel systems, and a 3D-printed high-throughput microreactor integrating 78 parallel zigzag channels is designed. 3D simulations reveal a three-stage mixing mechanism and uniform flow distribution among parallel channels, with total throughput two orders of magnitude higher than a single channel. Single-channel experiments with industrial germanium-bearing raffinate yield 91.81% precipitation efficiency under optimal conditions, reducing the reaction residence time from hours in conventional batch processes to the second scale. Staged reagent addition and two-stage serial configuration further raise the efficiency to ~98%, realizing deep germanium recovery with significantly improved reagent utilization and reduced impurity co-precipitation. This process achieves efficient intensification of the chelation precipitation process while balancing throughput and mixing performance, providing a novel and technically feasible approach for efficient low-consumption germanium recovery, and offering solid technical support for the industrial application of microreactors in the hydrometallurgy field. Full article

Central heating systems often rely on uniform regulation, making it difficult to meet the differentiated comfort demands of users and frequently leading to overheating. This paper proposes an end-to-end, data-driven framework that systematically couples density-adaptive clustering with dynamic response modeling for precision heating control. First, an adaptive DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm is developed, which automatically determines its parameters via k-distance graph initialization, differential evolution optimization, and hierarchical clustering post-processing. Without requiring a pre-set cluster number, it consistently identifies four typical daily room temperature fluctuation patterns. Validated on 120-day data from a residential community in Luoyang, the first four clusters cover over 80% of users, and the clustering quality approaches that of manually optimized conventional methods. Second, multi-input ARX (Autoregressive with Exogenous Inputs) models are built for the representative user of each cluster to characterize dynamic responses to supply water temperature, flow rate, and outdoor temperature. Rolling prediction for the entire community achieves an RMSE of 0.24 °C and an R² of 0.93. Finally, a differentiated regulation strategy combining main-cluster supply temperature control and small-cluster flow compensation is designed. Simulation results demonstrate that this strategy drives the room temperatures of all clusters significantly toward the 20 °C comfort target, with a marked reduction in standard deviation. The primary contribution of this study lies in the construction of a reproducible, closed-loop pipeline—from raw room temperature data to demand-based regulation logic—offering a quantitative basis for central heating systems transitioning from passive balancing to data-driven, classified control. Full article

Heavy and extra-heavy oils represent a large and growing share of recoverable hydrocarbon resources, yet their extreme viscosity, high heteroatom content, and non-Newtonian behavior routinely defeat empirical correlations developed for conventional crude. Machine learning has emerged as a candidate response to this modeling gap, but existing reviews largely catalog applications without asking whether the technology is actually ready for industrial deployment. This critical review synthesizes machine learning applications across five thematic domains of the heavy-oil value chain: physicochemical property prediction, enhanced oil recovery, flow assurance, reactive recovery, and downstream upgrading. Studies are read through a three-phase historical lens, tracing the field’s progression from empirical-correlation replacement to methodological diversification to physics-informed and closed-loop integration, and evaluated against a Technology Readiness Level (TRL) scale adapted specifically for heavy-oil machine learning. The multilayer perceptron anchors more of the primary corpus than any other architecture, a pattern that, in our interpretation, reflects small-sample, low-dimensional regression needs rather than any demonstrated advantage over other architectures. Enhanced oil recovery is the only cluster to reach organizational-scale deployment, anchored by a single multi-decade operator program, Chevron’s San Joaquin Valley i-field; the remaining clusters are constrained less by modeling sophistication than by single-basin datasets and undisclosed uncertainty. Measured against three falsifiable deployability criteria, fidelity preservation below 10° API, operator-grade interpretability, and demonstrated laboratory-to-field transferability, no study in the reviewed corpus is documented to satisfy all three simultaneously; because industrial implementations are frequently proprietary, this is a statement about the published record identified by this search, not a claim that the capability does not exist. Federated learning, physics-informed architectures, and sequence-aware models emerge as the directions most likely to close this gap. Full article

In the testing of aircraft cargo smoke detectors, surrogate smoke is often used in place of fire-generated smoke to avoid the hazards of live-fire experiments. Reproducing the time-varying concentration profile of real fire smoke requires feedback control of the surrogate smoke concentration. Two obstacles arise: the turbulent smoke flow is difficult to model accurately, and the distance between the generator and detector introduces a substantial control-loop delay. This study proposes a smoke simulation method based on model-free adaptive predictive control (MFAPC). The MFAPC scheme was tested in a full-scale aircraft cargo compartment simulator, where it drove the surrogate smoke concentration to track the profile recorded from a real cargo fire. Particle image velocimetry (PIV) was used concurrently with concentration control to capture the corresponding smoke velocity field. Across all conditions, MFAPC reduced the root-mean-square error by up to 38% compared with model-free adaptive control alone. With a control-loop delay longer than 10 s, the light transmission deviation remained within 2% of the target. The PIV data show that the controlled surrogate smoke velocity field reproduces the dominant structures and evolution patterns of actual fire-generated smoke, providing fluid-mechanistic evidence that a recreated dynamic smoke environment is physically meaningful. Full article

Shale gas reservoirs are characterized by low permeability, nanoscale pore structures, and complex fracture networks. Multistage fractured horizontal wells are an important technology for commercial shale gas development. However, many shale gas productivity models primarily emphasize gas transport within the reservoir and fracture system, while pressure variations caused by frictional losses along the horizontal wellbore are often simplified or treated separately. To address this issue, this study develops a fully coupled multiscale dual-porosity numerical model that integrates the shale matrix, hydraulic fractures, and horizontal wellbore within a unified simulation framework. The model incorporates key physical mechanisms governing shale gas transport, including Knudsen diffusion, Langmuir adsorption–desorption, stress sensitivity, and non-Darcy flow in fractures. Meanwhile, the Darcy–Weisbach equation is introduced to describe wellbore frictional pressure losses. The reliability of the proposed model is validated through history matching with field production data from the Changning shale gas reservoir. The results demonstrate that neglecting wellbore friction losses leads to a 30–50% overestimation of horizontal well productivity, indicating that wellbore friction has a significant impact on fracture flow distribution and productivity prediction. Furthermore, an exponent factor r is introduced to characterize and evaluate non-uniform fracture placement patterns. The results show that toe-dense fracture placement can increase cumulative gas production by approximately 37.8% compared with uniform fracture placement when r = 1.10, which yields the highest cumulative gas production among the tested cases. However, the additional production benefit becomes substantially smaller after the initial increase and remains relatively stable as r further increases. This study improves the understanding of friction-induced heel-to-toe effects and provides an effective numerical approach for productivity prediction and fracture placement design in shale gas horizontal wells. Full article

Intrusion Detection Systems (IDSs) based on traditional machine learning treat network flows as independent tabular samples, ignoring the relational and topological structure that characterizes modern distributed attacks. Graph Neural Networks (GNNs) overcome this limitation by modeling network topology, which in turn raise the need to make their predictions transparent. This work develops and compares traditional classifiers against a GNN-based IDS on the UNSW-NB15 dataset, for both binary and multiclass classification. A novel graph construction is proposed in which each node is an individual flow and edges are defined by temporal proximity through three complementary strategies (conversation chains and temporal k-NN by source and destination IP). Three GNN backbones—GraphSAGE, Graph Convolutional Network (GCN) and Graph Attention Network (GAT)—are trained under an identical, matched pipeline and a chronological, inductive evaluation protocol, so that any difference is attributable to the backbone alone. A two-stage classifier then separates detection from attack-type categorisation, with GNNExplainer providing interpretability, and SHAP applied to the traditional models. In binary classification, GraphSAGE achieves an Accuracy of 0.9906, Precision of 0.9856, Recall of 0.9998, F1-Score of 0.9927 and ROC-AUC of 0.9965, exceeding the traditional baselines in their conventional evaluation setting, while GCN and GAT reach comparable detection (F1 ≈ 0.99), showing that the temporal graph rather than the specific backbone drives detection. The explainability analysis identifies TTL-related and connection-state variables as dominant predictors and reveals attack-specific structural patterns, confirming that temporally structured GNNs improve detection while providing interpretable predictions. Full article