Search Results (48,799)

Background: Artificial intelligence (AI) is increasingly used in intensive care units (ICUs) to enable personalized care, real-time analytics, and decision support. Nutritional therapy—a major determinant of ICU outcomes—often remains delayed or non-individualized. Objective: This study aimed to review current and emerging AI applications in ICU nutrition, highlighting clinical potential, implementation barriers, and ethical considerations. Methods: A narrative review of English-language literature (January 2018–November 2025) searched in PubMed/MEDLINE, Scopus, and Web of Science, complemented by a pragmatic Google Scholar sweep and backward/forward citation tracking, was conducted. We focused on machine learning (ML), deep learning (DL), natural language processing (NLP), and reinforcement learning (RL) applications for energy/protein estimation, feeding tolerance prediction, complication prevention, and adaptive decision support in critical-care nutrition. Results: AI models can estimate energy/protein needs, optimize EN/PN initiation and composition, predict gastrointestinal (GI) intolerance and metabolic complications, and adapt therapy in real time. Reinforcement learning (RL) and multi-omics integration enable precision nutrition by leveraging longitudinal physiology and biomarker trajectories. Key barriers are data quality/standardization, interoperability, model interpretability, staff training, and governance (privacy, fairness, accountability). Conclusions: With high-quality data, robust oversight, and clinician education, AI can complement human expertise to deliver safer, more targeted ICU nutrition. Implementation should prioritize transparency, equity, and workflow integration. Full article

Accurate parameter estimation of photovoltaic (PV) models is fundamentally a challenging nonlinear optimization problem, characterized by strong nonlinearity, high dimensionality, and multiple local optima. These characteristics significantly hinder the convergence accuracy, stability, and efficiency of conventional metaheuristic algorithms when applied to PV parameter identification. Although the enterprise development (ED) optimization algorithm has shown promising performance in various optimization tasks, it still suffers from slow convergence, limited solution precision, and poor robustness in complex PV parameter estimation scenarios. To overcome these limitations, this paper proposes a multi-strategy enhanced enterprise development (MEED) optimization algorithm for high-precision PV model parameter estimation. In MEED, a hybrid initialization strategy combining chaotic mapping and adversarial learning is designed to enhance population diversity and improve the quality of initial solutions. Furthermore, a historical trend-guided position update mechanism is introduced to exploit accumulated search information and accelerate convergence toward the global optimum. In addition, a mirror-reflection boundary control strategy is employed to maintain population diversity and effectively prevent premature convergence. The proposed MEED algorithm is first evaluated on the IEEE CEC2017 benchmark suite, where it is compared with 11 state-of-the-art metaheuristic algorithms under 30-, 50-, and 100-dimensional settings. Quantitative experimental results demonstrate that MEED achieves superior solution accuracy, faster convergence speed, and stronger robustness, yielding lower mean fitness values and smaller standard deviations on the majority of test functions. Statistical analyses based on Wilcoxon rank-sum and Friedman tests further confirm the significant performance advantages of MEED. Moreover, MEED is applied to the parameter estimation of single-diode and double-diode PV models using real measurement data. The results show that MEED consistently attains lower root mean square error (RMSE) and integrated absolute error (IAE) than existing methods while exhibiting more stable convergence behavior. These findings demonstrate that MEED provides an efficient and reliable optimization framework for PV model parameter estimation and other complex engineering optimization problems. Full article

Moso bamboo, widely distributed in subtropical regions of China, plays an important role in forest management and carbon cycle research. However, accurate estimation of tree density and aboveground biomass (AGB) remains challenging due to the unique characteristics of Moso bamboo forests in their growth and stand structure. This research aims to develop a new procedure for bamboo tree density and AGB estimation based on UAV-LiDAR and sample plots from multiple sites through comparative analysis of the incorporation of two groups of variables—regular point cloud metrics (e.g., height, point density) and layered texture metrics—and three modeling methods—multiple linear regression (MLR), mixed-effects modeling (MEM), and hierarchical Bayesian modeling (HBM). The results showed that incorporating layered texture metrics with regular variables substantially improved the estimation accuracy of both tree density and AGB. Among these models, HBM achieved the highest predictive performance, yielding coefficient of determination (R²) values of 0.54 for tree density and 0.59 for AGB, with corresponding relative root mean square errors (rRMSE) of 21.46% and 17.97%. This study presents a novel and effective method for estimating Moso bamboo tree density and AGB using multi-site UAV-LiDAR and sample plots, offering a scientific basis for precise management and carbon stock assessment. Full article

The increasing complexity of plant disease manifestations, especially in cases of multiple simultaneous infections, poses significant challenges to sustainable agriculture. To address this issue, we introduce the Apple Leaf Mixed Disease Recognition (ALMDR) model, a novel multi-task learning approach specifically designed for identifying and quantifying mixed disease infections in apple leaves. ALMDR comprises four key modules: a Group Feature Pyramid Network (GFPN) for multi-scale feature extraction, a Multi-Label Classification Head (MLCH) for disease type prediction, a Leaf Segmentation Head (LSH), and a Lesion Segmentation Head (LeSH) for precise delineation of leaf and lesion areas. The GFPN enhances the traditional Feature Pyramid Network (FPN) through differential sampling and grouping strategies, significantly improving the capture of fine-grained disease characteristics. The MLCH enables simultaneous classification of multiple diseases on a single leaf, effectively addressing the mixed infection problem. The segmentation heads (LSH and LeSH) work in tandem to accurately isolate leaf and lesion regions, facilitating detailed analysis of disease patterns. Experimental results on the Plant Pathology 2021-FGVC8 dataset demonstrate ALMDR’s effectiveness, outperforming state-of-the-art methods across multiple tasks. Our model achieves high performance in multi-label classification (F1-score of 93.74%), detection and segmentation (mean Average Precision (mAP) of 51.32% and 45.50%, respectively), and disease severity estimation ($R^2$ = 0.9757). Additionally, the model maintains this accuracy while processing 6.25 frames per second, balancing performance with computational efficiency. ALMDR demonstrates potential for real-time disease management in apple orchards, with possible applications extending to other crops. Full article

The rapid expansion of information on the Internet has rendered recommender systems vital for mitigating information overload. However, existing recommendation models based on heterogeneous information networks (HINs) often face challenges such as data sparsity and insufficient semantic utilization. Therefore, we propose a multi-layer attention hybrid recommendation model based on heterogeneous information networks (MLAHRec). Compared to traditional HIN-based recommendation models, we design a progressive three-layer attention architecture of “collaborative-node-path.” Specifically, collaborative attention first enhances the direct interaction representation between users and items. Subsequently, node attention filters important neighbor information on the same meta-path. Finally, path attention adaptively fuses the semantics of multiple meta-paths, thereby achieving hierarchical refinement from micro-level interactions to macro-level semantics. Experiments on four real datasets, including MovieLens, LastFM, Yelp, and Douban-Movie, demonstrate that MLAHRec significantly outperforms mainstream baseline algorithms, as determined by Precision@10, Recall@10, and NDCG@10 metrics, validating the effectiveness and interpretability of the model. Full article

Utilizing waste tire crumb rubber to modify asphalt enhances the damping and noise reduction performance of pavements. This study employs a multi-scale approach to investigate the effect of crumb rubber content (5–25%) on the damping performance of crumb rubber-modified asphalt (CRMA). The results show that damping performance improves initially with increasing crumb rubber content, peaking at 20%, and then declines. At this optimal content, the loss modulus increases by 110% and 440% at 46 °C and 82 °C, respectively, compared to base asphalt, with enhanced damping efficiency and damping temperature stability. Fluorescence microscopy (FM) images and quantitative analysis reveal that, at 20%, the crumb rubber forms a moderately connected three-dimensional network. Molecular dynamics (MD) simulations indicate that, at this content, the solubility parameter of the CRMA system is closest to that of the base asphalt, and interfacial binding energy increases, suggesting optimal compatibility. Ridge regression models, with R² values of 0.903 and 0.876 for the FM and MD scales, respectively, confirm that crumb rubber dispersion is the dominant factor governing damping performance, with moderate phase separation further enhancing performance. This study establishes a quantitative structure–property relationship, providing a framework for understanding the damping performance of rubber-modified asphalt pavements. Full article

Liquid-fueled molten salt reactors (MSRs) are characterized by the use of liquid nuclear fuel, which leads to a unique thermal-hydraulic phenomenon in the core involving the simultaneous occurrence of nuclear fission heat generation and convective heat transfer. This distinctive behavior creates a critical need for high-fidelity experimental data on internally heated flows, yet such studies are severely constrained by the lack of methods to generate controllable, high-power-density volumetric heat sources in fluids. To address this methodological gap, this study proposes and numerically investigates a novel experimental concept based on microwave heating. The design features an innovative multi-tier hexagonal resonant cavity with fifteen strategically staggered magnetrons. A comprehensive multi-physics model was developed using COMSOL Multiphysics to simulate the coupled electromagnetic, thermal, and fluid flow processes. Simulation results confirm the feasibility of generating a volumetric heat source, achieving an average power density of 6.9 MW/m³. However, the inherent non-uniformity in microwave power deposition was quantitatively characterized, revealing a high coefficient of variation (COV) for power density. Crucially, parametric studies demonstrate that this non-uniformity can be effectively mitigated by optimizing the flow channel geometry. Specifically, using a smaller diameter tube or an annulus pipe significantly improved temperature field uniformity, reducing the temperature COV by over an order of magnitude, albeit at the cost of reduced absorption efficiency. Preliminary discussion also addresses the extension of this approach towards molten salt experiments. The findings establish a practical design framework and provide quantitative guidance for subsequent experimental investigations into the thermal-hydraulic behavior of internally heated fluids, offering a promising pathway to support the design and safety analysis of liquid-fueled MSRs. Full article

Amidst the accelerating trend of population aging, addressing regional disparities in basic elderly care services (BECS for short) and identifying the key obstacles to their development have become crucial prerequisites for development. Taking urgent transformation measures is indispensable for enhancing the quality of fundamental senior care provisions and advancing the attainment of the United Nations Sustainable Development Goals (SDGs for short) by 2030. However, the extant literature does not have a sufficient understanding of the evolution of differences, spatial correlations, and sources of obstacles. Therefore, this paper takes the period from 2021 to 2023 as the investigation period and comprehensively applies the entropy weight method, Dagum Gini coefficient, kernel density estimation, Moran Index, and obstacle degree model to conduct a systematic analysis of BECS in China. Quantitative results obtained from the research demonstrate that the level of BECS in China follows the pattern of eastern > western > central > northeastern regions. The overall difference slightly increases, and the differences within and between regions vary. The kernel density estimation results are highly consistent with the current landscape of the level of BECS in China, and the spatial correlation and aggregation characteristics are obvious. It was also found that the main obstacles in the quasi-measurement layer (including the indicator layer) were concentrated in the dimension of welfare subsidies. Based on this, a policy combination proposal is put forward in terms of strengthening the construction of a multi-subject supply network, promoting the cross-regional coordinated development of human, financial, and material factors, and enhancing the government’s governance capacity, with the aim of increasing Chinese contributions to improving the level of BECS and achieving the United Nations 2030 Sustainability Goals on schedule. Full article

The accurate determination of nuclear level density (NLD) is essential for a wide range of applications in nuclear science, including reactor design, nuclear astrophysics, and nuclear data evaluation. Traditional phenomenological models often face challenges in capturing key physical effects, such as collective excitations and shell structure, particularly in heavy and transitional nuclei, where the level density grows exponentially. Machine learning (ML) approaches have shown promise in improving predictive accuracy but are often limited by their purely data-driven nature, leading to challenges in interpretability and performance in regions with sparse experimental data. In this study, we propose a Physics-Informed Neural Network (PINN) framework, enhanced through multi-task learning (MTL), to address these limitations. The proposed model simultaneously predicts cumulative levels and mean resonance spacings by integrating experimental data with theoretical constraints, ensuring consistency with nuclear structure theory and robustness in extrapolating beyond the training data. Validation against both cumulative and yrast levels highlights the model’s ability to accurately capture rotational and vibrational excitations across a wide range of isotopes. Comparative evaluations demonstrate that the PINN model significantly outperforms traditional phenomenological models and purely data-driven approaches, offering a comprehensive and interpretable framework for advancing nuclear level density predictions and supporting practical applications in nuclear energy and astrophysics. Full article

In densely populated and traffic-congested major cities, medical emergency rescue incidents occur frequently, making the use of drones for emergency medical supplies delivery a new emergency distribution method. However, establishing drone transportation networks in urban areas requires balancing spatiotemporal fluctuations in emergency needs, meeting hospitals’ mandatory constraints on response time, and addressing factors like airspace restrictions and weather impacts. By analyzing the spatiotemporal distribution characteristics of medical emergency logistics in large cities, this study constructs a drone base station location optimization model integrating dynamic and static factors. The model combines multi-source data including emergency needs, geographic information, and airspace limitations. It employs kernel density estimation to identify hotspot areas, uses DBSCAN clustering to detect long-term stable demand hotspots, and applies LSTM methods to predict short-term and sudden demand fluctuations. The model optimizes coverage rate, response time, and cost budget control for drone transportation networks through a multi-objective genetic algorithm. Using Guangzhou as a case study, the results demonstrate that through “dynamic-static” collaborative deployment and multi-model drone coordination, the network achieves 96.18% demand coverage with an average response time of 673.38 s, significantly outperforming traditional vehicle transportation. Sensitivity analysis and robustness testing further validate the model’s effectiveness in handling demand fluctuations, weather changes, and airspace restrictions. This research provides theoretical support and decision-making basis for scientific planning of urban medical emergency drone transportation networks, offering practical significance for enhancing urban emergency rescue capabilities. Full article

The increasing share of renewable energy sources (RES) in modern power systems necessitates the development of efficient, large-scale energy storage technologies capable of mitigating generation variability. Liquid Air Energy Storage (LAES), particularly in its adiabatic form, has emerged as a promising candidate by leveraging thermal energy storage and high-pressure air liquefaction and regasification processes. Although LAES has been widely studied, the impact of ambient temperature on its performance remains insufficiently explored. This study addresses that gap by examining the thermodynamic response of an adiabatic LAES system under varying ambient air temperatures, ranging from 0 °C to 35 °C. A detailed mathematical model was developed and implemented in Aspen Hysys to simulate the system, incorporating dual refrigeration loops (methanol and propane), thermal oil intercooling, and multi-stage compression/expansion. Simulations were conducted for a reference charging power of 42.4 MW at 15 °C. The influence of external temperature was evaluated on key parameters including mass flow rate, unit energy consumption during liquefaction, energy recovery during expansion, and round-trip efficiency. Results indicate that ambient temperature has a marginal effect on overall LAES performance. Round-trip efficiency varied by only ±0.1% across the temperature spectrum, remaining around 58.3%. Mass flow rates and power output varied slightly, with changes in discharging power attributed to temperature-driven improvements in expansion process efficiency. These findings suggest that LAES installations can operate reliably across diverse climate zones with negligible performance loss, reinforcing their suitability for global deployment in grid-scale energy storage applications. Full article

Deep learning for semantic segmentation has made significant advances in recent years, achieving state-of-the-art performance. Medical image segmentation, as a key component of healthcare systems, plays a vital role in the diagnosis and treatment planning of diseases. Due to the fractal and scale-invariant nature of biological structures, effective medical image segmentation requires models capable of capturing hierarchical and self-similar representations across multiple spatial scales. In this paper, a Recurrent Neural Network (RNN) is explored within the Convolutional Neural Network (CNN) and Vision Transformer (ViT)-based hybrid U-shape network, named RCV-UNet. First, the ViT-based layer was developed in the bottleneck to effectively capture the global context of an image and establish long-range dependencies through the self-attention mechanism. Second, recurrent residual convolutional blocks (RRCBs) were introduced in both the encoder and decoder to enhance the ability to capture local features and preserve fine details. Third, by integrating the global feature extraction capability of ViT with the local feature enhancement strength of RRCBs, RCV-UNet achieved promising global consistency and boundary refinement, addressing key challenges in medical image segmentation. From a fractal–fractional perspective, the multi-scale encoder–decoder hierarchy and attention-driven aggregation in RCV-UNet naturally accommodate fractal-like, scale-invariant regularity, while the recurrent and residual connections approximate fractional-order dynamics in feature propagation, enabling continuous and memory-aware representation learning. The proposed RCV-UNet was evaluated on four different modalities of images, including CT, MRI, Dermoscopy, and ultrasound, using the Synapse, ACDC, ISIC 2018, and BUSI datasets. Experimental results demonstrate that RCV-UNet outperforms other popular baseline methods, achieving strong performance across different segmentation tasks. The code of the proposed method will be made publicly available. Full article

Fuel cell commercial vehicles are widely used in commercial transport for their high efficiency and long range. However, in mixed operating scenarios, their energy economy and fuel cell operational stability cannot be fully balanced. Traditional strategies lack adaptability in mixed operating scenarios. Therefore, based on the equivalent factor regulation formula of the Adaptive Equivalent Hydrogen Consumption Minimization Strategy (A-ECMS) and the improved Sparrow Search Algorithm-Long Short-Term Memory (SSA-LSTM) hybrid model, short-term speed prediction and three-stage speed interval division are embedded into the equivalent factor regulation logic. A dynamic equivalent factor regulation strategy integrating SOC deviation is constructed, and an improved Predictive Equivalent Hydrogen Consumption Minimization Strategy (P-ECMS) is finally derived. The SSA-LSTM algorithm is optimized via constrained hyperparameter tuning for short-term speed prediction. A time-decay weighting mechanism enhances recent speed data weight, with weighted results as inputs to boost accuracy. Moving Average Residual Correction (MARC) is used to verify the speed prediction model accuracy and correct residuals. Multi-scenario tests show that the SSA-LSTM model outperforms the Gated Recurrent Unit (GRU) model in prediction accuracy and generalization ability, providing reliable data support for segmented regulation. With battery SOC deviation and the SSA-LSTM-predicted speed trend as core inputs, combined with three-stage speed interval division, A-ECMS’s equivalent factor regulation formula is improved. The model adopts a segmented dynamic regulation logic to integrate dual factors into equivalent factor adjustment, and it reasonably adjusts the energy output ratio of fuel cells and power batteries according to speed intervals and operating condition changes. In scenarios with significant speed fluctuations and frequent operating condition transitions, power shocks are mitigated by the power battery’s peak-shaving and valley-filling function. Simulation results for C-WTVC and NREL2VAIL show that, compared with traditional A-ECMS, the improved P-ECMS has notable energy benefits, with equivalent hydrogen consumption reduced by 3.41% and 5.48%, respectively. The fuel cell’s state is significantly improved, with its high-efficiency share reaching 63%. The output power curve is smoother, start–stop losses are reduced, and the fuel cell’s service life is extended, balancing the energy economy and component durability. Full article

Small object detection in remote sensing images provides significant value for urban monitoring, aerospace reconnaissance, and other fields. However, detection accuracy still faces multiple challenges including limited target information, weak feature representation, and complex backgrounds. This research aims to improve the performance of the YOLO11 model for small object detection in remote sensing imagery by addressing key issues in long-distance spatial dependency modeling, multi-scale feature adaptive fusion, and computational efficiency. We constructed a specialized Remote Sensing Airport-Plane Detection (RS-APD) dataset and used the public VisDrone2019 dataset for generalization verification. Based on the YOLO11 architecture, we proposed the DAE-YOLO model with three innovative modules: Dynamic Spatial Sequence Module (DSSM) for enhanced long-distance spatial dependency capture; Adaptive Multi-scale Feature Enhancement (AMFE) for multi-scale feature adaptive receptive field adjustment; and Efficient Dual-level Attention Mechanism (EDAM) to reduce computational complexity while maintaining feature expression capability. Experimental results demonstrate that compared to the baseline YOLO11, our proposed model improved mAP₅₀ and mAP_50:95 on the RS-APD dataset by 2.1% and 2.5%, respectively, with AP_s increasing by 2.8%. This research provides an efficient and reliable small object detection solution for remote sensing applications. Full article

Explosive volcanic eruptions are key drivers of climate variability, yet their hemispheric-dependent impacts remain uncertain. Using multi-model ensembles from Coupled Model Intercomparison Project Phase 6 (CMIP6) historical data and Decadal Climate Prediction Project (DCPP) simulations, this study examines how the spatial distribution of volcanic aerosols modulates climate responses to Northern Hemisphere (NH), Tropical (TR), and Southern Hemisphere (SH) eruptions. The CMIP6 ensemble captures observed temperature and precipitation patterns, providing a robust basis for assessing volcanic effects. The results show that the hemispheric distribution of aerosols strongly controls radiative forcing, surface air temperature, and hydrological responses. TR eruptions cause nearly symmetric cooling and widespread tropical rainfall reduction, while NH and SH eruptions produce asymmetric temperature anomalies and clear Intertropical Convergence Zone (ITCZ) displacements away from the perturbed hemisphere. The vertical temperature structure, characterized by stratospheric warming and tropospheric cooling, further amplifies hemispheric contrasts through enhanced cross-equatorial energy transport and shifts in the Hadley circulation. ENSO-like responses depend on eruption latitude, TR and NH eruptions favor El Niño–like warming through westerly wind anomalies and Bjerknes feedback, and SH eruptions induce La Niña–like cooling. The DCPP experiments confirm that these signals primarily arise from volcanic forcing rather than internal variability. These findings highlight the critical role of aerosol asymmetry and vertical temperature structure in shaping post-eruption climate patterns and advancing the understanding of volcanic–climate interactions. Full article