Search Results (689)

Accurate crop type mapping remains challenging in regions where persistent cloud cover limits the availability of optical imagery. Multi-temporal dual-polarization Sentinel-1 SAR data offer an all-weather alternative, yet existing approaches often underutilize polarization information and rely on single-scale temporal aggregation. This study proposes PTU-Net, a polarization–temporal U-Net designed specifically for pixel-wise crop segmentation from SAR time series. The model introduces a Polarization Channel Attention module to construct physically meaningful VV/VH combinations and adaptively enhance their contributions. It also incorporates a Multi-Scale Temporal Self-Attention mechanism to model pixel-level backscatter trajectories across multiple spatial resolutions. Using a 12-date Sentinel-1 stack over Kings County, California, and high-quality crop-type reference labels, the model was trained and evaluated under a spatially independent split. Results show that PTU-Net outperforms GRU, ConvLSTM, 3D U-Net, and U-Net–ConvLSTM baselines, achieving the highest overall accuracy and mean IoU among all tested models. Ablation studies confirm that both polarization enhancement and multi-scale temporal modeling contribute substantially to performance gains. These findings demonstrate that integrating polarization-aware feature construction with scale-adaptive temporal reasoning can substantially improve the effectiveness of SAR-based crop mapping, offering a promising direction for operational agricultural monitoring. Full article

With the increasing penetration of renewable energy sources and power-electronic devices, modern power systems exhibit pronounced wideband oscillation characteristics with large frequency spans, strong modal coupling, and significant time-varying behaviors. Accurate identification and classification of wideband oscillation patterns have therefore become critical challenges for ensuring the secure and stable operation of “dual-high” power systems. Existing methods based on signal processing or single-modality deep-learning models often fail to fully exploit the complementary information embedded in heterogeneous data representations, resulting in limited performance when dealing with complex oscillation patterns.To address these challenges, this paper proposes a multimodal attention-based fusion network for wideband oscillation classification. A dual-branch deep-learning architecture is developed to process Gramian Angular Difference Field images and raw time-series signals in parallel, enabling collaborative extraction of global structural features and local temporal dynamics. An improved Inception module is employed in the image branch to enhance multi-scale spatial feature representation, while a gated recurrent unit network is utilized in the time-series branch to model dynamic evolution characteristics. Furthermore, an attention-based fusion mechanism is introduced to adaptively learn the relative importance of different modalities and perform dynamic feature aggregation. Extensive experiments are conducted using a dataset constructed from mathematical models and engineering-oriented simulations. Comparative studies and ablation studies demonstrate that the proposed method significantly outperforms conventional signal-processing-based approaches and single-modality deep-learning models in terms of classification accuracy, robustness, and generalization capability. The results confirm the effectiveness of multimodal feature fusion and attention mechanisms for accurate wideband oscillation classification, providing a promising solution for advanced power system monitoring and analysis. Full article

High-accuracy short-term electric load forecasting is essential for ensuring the security of power systems and enhancing energy efficiency. Power load sequences are characterized by strong randomness, non-stationarity, and nonlinearity over time. To improve the precision and efficiency of short-term load forecasting in microgrids, a hybrid predictive model combining Complementary Ensemble Empirical Mode Decomposition (CEEMD) and a multi-strategy enhanced Whale Optimization Algorithm (WOA) with Long Short-Term Memory (LSTM) neural networks has been proposed. Initially, this study employs CEEMD to decompose the short-term electric load time series. Subsequently, a multi-strategy enhanced WOA with chaotic initialization and reverse learning is introduced to enhance the search capability of model parameters and avoid entrapment in local optima. Finally, considering the distinct characteristics of each component, the multi-strategy improved WOA is utilized to optimize the LSTM model, establishing individual predictive models for each component, and the predictions are then aggregated. The proposed method’s forecasting accuracy has been validated through multiple case studies using the UC San Diego microgrid data, demonstrating its reliability and providing a solid foundation for microgrid system planning and stable operation. Full article

Background: Cultivated oral mucosal epithelial transplantation (COMET/CAOMECS) is an autologous, immunosuppression-sparing option for ocular surface reconstruction in limbal stem cell deficiency (LSCD). After two decades, indications, platforms and outcome definitions vary, and COMET’s position relative to limbal-derived epithelium remains uncertain. Methods: We conducted a PRISMA-ScR scoping review of human clinical studies (PubMed, 2000–30 December 2025) with hand-searching and regulatory sources. Eligible reports included COMET/CAOMECS series and comparative cohorts (CLET/ACLET, SLET, KLAL/CLAL). The primary outcome was anatomical success (stable epithelialised cornea without recurrent persistent epithelial defect, progressive conjunctivalisation or uncontrolled neovascularisation at last assessment). Given heterogeneity in definitions and analytic frames (fixed-time vs. Kaplan–Meier [KM]), results were synthesised narratively by indication and platform. Results: Twenty-five reports (893 eyes; 821 patients) were included. Aetiologies were predominantly burns and SJS/TEN. Across amniotic membrane-based mixed-aetiology series, 12-month anatomical success clustered around 55–70%. Aggregated descriptively across COMET eyes, 211/467 (45%) had a stable surface at last follow-up. Epithelialisation was generally rapid in quiet AM-based reconstructions and slower with severe adnexal disease or carrier-free platforms. Mean BCVA improved from 1.8 ± 0.7 to 1.4 ± 0.7 logMAR (471 eyes); ≥2-line gains occurred in 308/471 (65.4%). A matched comparison suggested better 12-month survival, less neovascularisation and better BCVA with substrate-free versus AM-carried COMET; a biomaterial-/feeder-free platform reconstructed most eyes but failed more often with four-quadrant symblepharon. Observational comparative cohorts suggested higher surface survival and average visual gain with limbal-derived epithelium, at the cost of systemic immunosuppression. Conclusions: In appropriately selected bilateral LSCD, COMET offers immunosuppression-sparing reconstruction with moderate, durable surface stability and clinically meaningful visual gains when performed on a quiet, optimised surface. Platform refinements—particularly substrate-free constructs—and prospective, indication-defined comparative studies with harmonised outcomes are needed to define COMET’s role relative to limbal-derived epithelium. Full article

Background/Objectives: Valve replacement (VR) patients are at high risk of postoperative complications, but reproducible Electronic Health Record (EHR) benchmarks for evaluating sequential AI models in this setting are lacking. We develop a reproducible pipeline that extracts two EHR datasets from MIMIC-IV (a general-purpose and a predictive benchmark dataset) capturing perioperative histories, high-resolution time-series, and clinically motivated outcome labels. Methods: The cohort comprises 3890 VR patients with clinician-guided feature selection across diagnoses, procedures, laboratory measurements, medications, and physiological monitoring. As an exemplary use case, we define ICU readmission at first ICU discharge as a surrogate for postoperative risk and derive a predictive benchmark under strict label-leakage control. We then compare a Transformer model trained on tokenized longitudinal EHR sequences with Transformer and XGBoost baselines trained on aggregated feature statistics, and assess performance differences using paired statistical tests across validation splits. Results: ICU readmission stratified in-hospital and 100-day outcomes, including mortality, complications, and rehospitalization, confirming the clinical relevance of the prediction target. The sequential Transformer achieved 0.87 AUROC and 0.69 AUPRC. Corrected resampled t-tests confirm improved performance over the non-sequential Transformer, while the comparison with XGBoost indicates a favorable trend without conclusive evidence. Conclusions: Our findings suggest that leveraging longitudinal EHR sequences yields higher predictive performance than static feature summaries for postoperative risk prediction. The publicly released preprocessing pipeline and cohort-construction code enable researchers with MIMIC-IV access to reproduce the datasets and provide a robust benchmark for developing and comparing time-series models in post-valve replacement care. Full article

Accurate forecasting of citywide electric vehicle (EV) charging load is critical for alleviating station-level congestion, improving energy dispatching, and supporting the stability of intelligent transportation systems. However, large-scale EV charging networks exhibit complex and heterogeneous spatio-temporal dependencies, and existing approaches often struggle to scale with increasing station density or long forecasting horizons. To address these challenges, we develop a modular spatio-temporal prediction framework that decouples temporal sequence modeling from spatial dependency learning under an encoder–decoder paradigm. For temporal representation, we introduce a global aggregation mechanism that compresses multi-station time-series signals into a shared latent context, enabling efficient modeling of long-range interactions while mitigating the computational burden of cross-channel correlation learning. For spatial representation, we design a dynamic multi-scale attention module that integrates graph topology with data-driven neighbor selection, allowing the model to adaptively capture both localized charging dynamics and broader regional propagation patterns. In addition, a cross-step transition bridge and a gated fusion unit are incorporated to improve stability in multi-horizon forecasting. The cross-step transition bridge maps historical information to future time steps, reducing error propagation. The gated fusion unit adaptively merges the temporal and spatial features, dynamically adjusting their contributions based on the forecast horizon, ensuring effective balance between the two and enhancing prediction accuracy across multiple time steps. Extensive experiments on a real-world dataset of 18,061 charging piles in Shenzhen demonstrate that the proposed framework achieves superior performance over state-of-the-art baselines in terms of MAE, RMSE, and MAPE. Ablation and sensitivity analyses verify the effectiveness of each module, while efficiency evaluations indicate significantly reduced computational overhead compared with existing attention-based spatio-temporal models. Full article

An extraordinary episode is a torrential rainfall event that produces significant societal impacts, which poses a major natural hazard in the western Mediterranean, particularly along the Valencia coast. This study evaluates the feasibility and added value of an explicitly spatial approach for estimating return periods of extraordinary precipitation in the Júcar and Turia basins, moving beyond traditional point-based or micro-catchment analyses. Our methodology consists of progressive spatial aggregation of time series within a basin to better estimate return periods of exceeding specific catastrophic rainfall thresholds. This technique allows us to compare 10 min rainfall data of a reference station (e.g., Turís, València, 29 October 2024 catastrophe) with long-term annual maxima from 98 stations. Temporal structure is characterized using the fractal–intermittency n-index, while tail behavior is modeled using several extreme-value distributions (Gumbel, GEV, Weibull, Gamma, and Pareto) and guided by empirical errors. Results show that $n\approx 0.3$–0.4 is consistent for extreme rainfall, while return periods systematically decrease as stations are added, stabilizing with about 15–20 stations, once the relevant spatial heterogeneity is sampled. Specifically, the probability of exceeding extraordinary thresholds is between 3 and 10 times higher for the areal than the point approach, so recurrence of a catastrophe would be once a few decades rather than centuries. Overall, the results demonstrate that spatially integrated return-period estimation is operational, physically consistent, and better suited for basin-scale risk assessment than purely point-based approaches, providing a relevant baseline for interpreting recent catastrophic events in the context of ongoing climatic warming in the Mediterranean region. Full article

Considering the nonlinear trends, multi-scale variations, and capacity regeneration phenomena exhibited by battery capacity degradation under real-world conditions, accurately predicting its trajectory remains a critical challenge for ensuring the reliability and safety of electric vehicles. To address this, this study proposes a hybrid prediction framework based on Variational Mode Decomposition and a Transformer–Long Short-Term Memory architecture. Specifically, the proposed Variational Mode Decomposition–Transformer for Time Series–Long Short-Term Memory (VMD–TTS–LSTM) framework first decomposes the capacity sequence using Variational Mode Decomposition. The resulting modal components are then aggregated into high-frequency and low-frequency parts based on their frequency centroids, followed by targeted feature analysis for each part. Subsequently, a simplified Transformer encoder (Transformer for Time Series, TTS) is employed to model high-frequency fluctuations, while a Long Short-Term Memory (LSTM) network captures the long-term degradation trends. Evaluated on charging data from 20 commercial electric vehicles under a long-horizon setting of 20 input steps predicting 100 steps ahead, the proposed method achieves a mean absolute error of 0.9247 and a root mean square error of 1.0151, demonstrating improved accuracy and robustness. The results confirm that the proposed frequency-partitioned, heterogeneous modeling strategy provides a practical and effective solution for battery health prediction and energy management in real-world electric vehicle operation. Full article

As a fundamental mathematical model for complex systems, complex networks describe interactions among social, infrastructural, and biological systems. However, the complete connection structure is often unobservable, making topology reconstruction from limited data—such as time series of unit states—a crucial challenge. To address network reconstruction under sparse local observations, this paper proposes a novel framework that integrates epidemic dynamics with Dempster–Shafer (DS) evidence theory. The core of our method lies in a two-level belief fusion process: (1) Intra-node fusion, which aggregates multiple independent SIR simulation results from a single seed node to generate robust local evidence represented as Basic Probability Assignments (BPAs), effectively quantifying uncertainty; (2) Inter-node fusion, which orthogonally combines BPAs from multiple seed nodes using DS theory to synthesize a globally consistent network topology. This dual-fusion design enables the framework to handle uncertainty and conflict inherent in sparse, stochastic observations. Extensive experiments demonstrate the effectiveness and robustness of the proposed approach. It achieves stable and high reconstruction accuracy on both a synthetic 16-node benchmark network and the real-world Zachary’s Karate Club network. Furthermore, the method scales successfully to four large-scale real-world networks, attaining an average accuracy of 0.85, thereby confirming its practical applicability across networks of different scales and densities. Full article

Hemostasis depends on the coordinated interaction between platelets, coagulation factors, endothelium, and shear forces. Current point-of-care (POC) assays evaluate isolated components of haemostasis or operate under nearly static conditions, limiting their ability to reproduce physiological thrombus formation. In this study, we performed the technical validation of Smart Clot, a fully automated, microfluidic POC platform that quantifies platelet aggregation, fibrin formation, and total thrombus growth under controlled arterial shear using unmodified whole blood. Recalcified citrated blood was perfused over collagen at $\dot{\gamma }$_w = 300 s⁻¹. Dual-channel epifluorescence microscopy acquired platelet and fibrin(ogen) signals at 1 frame per second. Integrated Density time-series were fitted with a five-parameter logistic model; first derivatives and their integrals yielded standardized pseudo-volumes for platelets, fibrin(ogen), and total thrombus. Sixty-two healthy donors established reference distributions; one-hundred-thirteen patients on antithrombotic therapy assessed pharmacodynamic sensitivity. Platelet-derived parameters were approximately normally distributed, whereas fibrin(ogen) and total thrombus values followed log-normal patterns. Anticoagulants and antiplatelet agents produced graded, mechanism-consistent inhibition of all thrombus components. Cardiopulmonary bypass samples showed profound but transient suppression of platelet and fibrin activity. Across activity ranges, multiple statistical assessments supported high analytical repeatability. Smart Clot provides rapid, reproducible, flow-aware quantification of platelet–fibrin dynamics, capturing pharmacological modulation and peri-operative impairment with high sensitivity. These results support its potential as a next-generation POC assay for physiological hemostasis assessment. Full article

The expansion of renewables in modern power systems and the coordinated development of upstream and downstream industrial chains are promoting a shift on the utility side from traditional settlement by energy toward operation driven by data and models. Industrial electricity consumption data exhibit pronounced multi-scale temporal structures and sectoral heterogeneity, which makes unified long-term load and generation forecasting while maintaining accuracy, interpretability, and scalability a challenge. From a modern system identification perspective, this paper proposes a System Identification in Adaptive Bayesian Framework (SIABF) for medium- and long-term industrial load forecasting based on daily freeze electricity time series. By combining daily aggregation of high-frequency data, frequency domain analysis, sparse identification, and long-term extrapolation, we first construct daily freeze series from 15 min measurements, and then we apply discrete Fourier transforms and a spectral complexity index to extract dominant periodic components and build an interpretable sinusoidal basis library. A sparse regression formulation with ${\ell }_1$ regularization is employed to select a compact set of key basis functions, yielding concise representations of sector and enterprise load profiles and naturally supporting multivariate and joint multi-sector modeling. Building on this structure, we implement a state-space-implicit physics-informed Bayesian forecasting model and evaluate it on real data from three representative sectors, namely, steel, photovoltaics, and chemical, using one year of 15 min measurements. Under a one-month-ahead evaluation using one year of 15 min measurements, the proposed framework attains a Mean Absolute Percentage Error (MAPE) of 4.5% for a representative PV-related customer case and achieves low single-digit MAPE for high-inertia sectors, often outperforming classical statistical models, sparse learning baselines, and deep learning architectures. These results should be interpreted as indicative given the limited time span and sample size, and broader multi-year, population-level validation is warranted. Full article

This study conducts a landslide deformation assessment in Tianshui, Gansu Province, on the Chinese Loess Plateau, utilizing the Small Baseline Subset InSAR (SBAS-InSAR) method integrated with velocity direction conversion and Z-score clustering. The Chinese Loess Plateau is one of the most landslide-prone regions in China due to frequent rains, strong topographical gradients and severe soil erosion. By constructing subsets of interferograms, SBAS-InSAR can mitigate the influence of decorrelation to a certain extent, making it a highly effective technique for monitoring regional surface deformation and identifying landslides. To overcome the limitations of the satellite’s one-dimensional Line-of-Sight (LOS) measurements and the challenge of distinguishing true landslide signals from noise, two optimization strategies were implemented. First, LOS velocities were projected onto the local steepest slope direction, assuming translational movement parallel to the slope. Second, a Z-score clustering algorithm was employed to aggregate measurement points with consistent kinematic signatures, enhancing identification robustness, with a slight trade-off in spatial completeness. Based on 205 Sentinel-1 Single-Look Complex (SLC) images acquired from 2014 to 2024, the integrated workflow identified 69 “active, very slow” and 63 “active, extremely slow” landslides. These results were validated through high-resolution historical optical imagery. Time series analysis reveals that creep deformation in this region is highly sensitive to seasonal rainfall patterns. This study demonstrates that the SBAS-InSAR post-processing framework provides a cost-effective, millimeter-scale solution for updating landslide inventories and supporting regional risk management and early warning systems in loess-covered terrains, with the exception of densely forested areas. Full article

Background/Objectives: Infection prevention and control (IPC) programs combine pathogen-targeted measures (e.g., admission screening) with hospital-wide standard precautions (e.g., hand hygiene, HH). We assessed temporal associations between screening, HH, antimicrobial stewardship (AMS), and hospital-level outcomes in a tertiary paediatric hospital. Methods: This study was a retrospective hospital-wide ecological time-series at the Children’s Memorial Health Institute. Annual aggregate data: 2011–2024 for screening, colonisation, and healthcare-associated infections (HAIs) with alert pathogens; 2016–2024 for antibiotic consumption (ATC J01, systemic antibacterials). Process indicators: number of screening tests and alcohol-based hand rub (ABHR) consumption per 1000 patient-days (PD). Outcomes: colonisations/HAIs per 1000 PD and defined daily doses (DDD) per 1000 PD overall and by class. Trends used linear regression and Spearman’s rank correlation. Results: Screening intensity increased from 39 to 150/1000 PD (slope +8.3/year; R² = 0.90; p < 0.001). Detected colonisation rose (2.5 → peak 8.05/1000 PD in 2023; slope +0.39; R² = 0.81; p < 0.001), while multidrug-resistant-organism (MDRO)-attributable HAIs remained low/stable (0.27–0.62/1000 PD; slope −0.014; p = 0.023). ABHR consumption increased from 26.1 to 78.0 L/1000 PD in 2020 (p < 0.001) and partially normalised to 60.0 in 2024 (>2 × baseline). Overall ATC J01 showed no long-term linear trend (~278–356 DDD/1000 PD; +2.57/year; p = 0.46), but class mix shifted: carbapenems, fluoroquinolones, and amoxicillin–clavulanate decreased; third/fourth-generation cephalosporins, piperacillin/tazobactam, and glycopeptides increased. Conclusions: In this tertiary paediatric setting, expansion of risk-based admission screening and sustained implementation of horizontal IPC measures were accompanied by increased detection of colonisation with alert pathogens, while MDRO-attributable HAIs remained low and stable at the hospital level. Over the same period, AMS activity coincided with a redistribution in antibiotic class use without a clear long-term reduction in total antibiotic consumption. These hospital-level findings are descriptive and hypothesis-generating; causal inference is limited by the ecological study design, and the heterogeneous, multispecialty structure of a tertiary paediatric centre. Full article

We present a new method for mapping global gas flaring using a multiyear spatio-temporal database of VIIRS Nightfire (VNF) nighttime infrared detections from the Suomi NPP, NOAA-20, and NOAA-21 satellites. The method is designed to resolve closely spaced industrial combustion sources and to produce a stable, physically meaningful flare catalog suitable for long-term monitoring and emissions analysis. The method combines adaptive spatial aggregation of high-temperature detections with a hierarchical clustering that super-resolves individual flare stacks within oil and gas fields. Post-processing yields physically consistent flare footprints and attraction regions, allowing separation of closely spaced sources. Flare clusters are assigned to operational categories (e.g., upstream, midstream, LNG) using prior catalogs combined with AI-assisted expert interpretation. In this step, a multimodal large language model (LLM) provides contextual classification suggestions based on geospatial information, high-resolution daytime imagery, and detection time-series summaries, while final attribution is performed and validated by domain experts. Compared with annual flare catalogs commonly used for national flaring estimates, the new catalog demonstrates substantially improved performance. It is more selective in the presence of intense atmospheric glow from large flares, identifies approximately twice as many active flares, and localizes individual stacks with ~50 m precision, resolving emitters separated by ~400–700 m. For the well-defined class of downstream flares at LNG export facilities, the catalog achieves complete detectability. These improvements support more accurate flare inventories, facility-level attribution, and policy-relevant assessments of gas flaring activity. Full article

In the domain of algorithmic decision-making, non-Markovian dynamics manifest as a significant impediment, especially for paradigms such as Reinforcement Learning (RL), thereby exerting far-reaching consequences on the advancement and effectiveness of the associated systems. Nevertheless, the existing benchmarks are deficient in comprehensively assessing the capacity of decision algorithms to handle non-Markovian dynamics. To address this deficiency, we have devised a generalized methodology grounded in category theory. Notably, we established the category of Markov Decision Processes (MDP) and the category of non-Markovian Decision Processes (NMDP), and proved the equivalence relationship between them. This theoretical foundation provides a novel perspective for understanding and addressing non-Markovian dynamics. We further introduced non-Markovianity into decision-making problem settings via the History Aggregator for State (HAS). With HAS, we can precisely control the state dependency structure of decision-making problems in the time series. Our analysis demonstrates the effectiveness of our method in representing a broad range of non-Markovian dynamics. This approach facilitates a more rigorous and flexible evaluation of decision algorithms by testing them in problem settings where non-Markovian dynamics are explicitly constructed. Full article