Search Results (1,446)

As large language models (LLMs) increasingly enter specialized domains, inference without external resources often leads to knowledge gaps, opaque reasoning, and hallucinations. To address these challenges in power equipment defect grading, we propose a zero-shot question-answering framework that requires no task-specific examples. Our system performs two-stage retrieval—first using a Sentence-BERT model fine-tuned on power equipment maintenance texts for coarse filtering, then combining TF-IDF and semantic re-ranking for fine-grained selection of the most relevant knowledge snippets. We embed both the user query and the retrieved evidence into a Chain-of-Thought (CoT) prompt, guiding the pre-trained LLM through multi-step reasoning with self-validation and without any model fine-tuning. Experimental results show that on a held-out test set of 218 inspection records, our method achieves a grading accuracy of 54.2%, which is 6.0 percentage points higher than the fine-tuned BERT baseline at 48.2%; an Explanation Coherence Score (ECS) of 4.2 compared to 3.1 for the baseline; a mean retrieval latency of 28.3 ms; and an average LLM inference time of 5.46 s. Ablation and sensitivity analyses demonstrate that a fine-stage retrieval pool size of k = 30 offers the optimal trade-off between accuracy and latency; human expert evaluation by six senior engineers yields average Usefulness and Trustworthiness scores of 4.1 and 4.3, respectively. Case studies across representative defect scenarios further highlight the system’s robust zero-shot performance. Full article

The increasing need for sustainable valorization of fossil-based and waste-derived materials has gained interest in converting complex organic matrices such as kerogen into valuable chemicals. This study explores a two-step oxidative strategy to decompose and valorize kerogen-rich oil shale, aiming to develop a locally based source of aliphatic dicarboxylic acids (DCAs). The method combines air oxidation with subsequent nitric acid treatment to enable selective breakdown of the organic structure under milder conditions. Air oxidation was conducted at 165–175 °C using 1% KOH as an alkaline promoter and 40 bar oxygen pressure (or alternatively 185 °C at 30 bar), targeting 30–40% carbon conversion. The resulting material was then subjected to nitric acid oxidation using an 8% HNO₃ solution. This approach yielded up to 23% DCAs, with pre-oxidation allowing a twofold reduction in acid dosage while maintaining efficiency. However, two-step oxidation was still accompanied by substantial degradation of the structure, resulting in elevated CO₂ formation, highlighting the need to balance conversion and carbon retention. The process offers a possible route for transforming solid fossil residues into useful chemical precursors and supports the advancement of regionally sourced, sustainable DCA production from unconventional raw materials. Full article

Background/Objectives: Heart rate variability (HRV) has been widely investigated as a predictor of disease and mortality across diverse patient populations; however, there remains no consensus on the optimal set or combination of time and frequency domain nor on nonlinear features for reliable prediction across clinical contexts. Given the relevance of the COVID-19 pandemic and the unique clinical profiles of these patients, this retrospective observational study explored the potential of HRV analysis for early prediction of in-hospital mortality using ECG signals recorded during the initial moments of ICU admission in COVID-19 patients. Methods: HRV indices were extracted from four ECG leads (I, II, III, and aVF) using sliding windows of 2, 5, and 7 min across observation intervals of 15, 30, and 60 min. The raw data posed significant challenges in terms of structure, synchronization, and signal quality; thus, from an original set of 381 records from 321 patients, after data pre-processing steps, a final dataset of 82 patients was selected for analysis. To manage data complexity and evaluate predictive performance, two feature selection methods, four feature reduction techniques, and five classification models were applied to identify the optimal approach. Results: Among the feature aggregation methods, compiling feature means across patient windows (Method D) yielded the best results, particularly for longer observation intervals (e.g., using LDA, the best AUC of $0.82\pm 0.13$ was obtained with Method D versus $0.63\pm 0.09$ with Method C using 5 min windows). Linear Discriminant Analysis (LDA) was the most consistent classification algorithm, demonstrating robust performance across various time windows and further improvement with dimensionality reduction. Although Gradient Boosting and Random Forest also achieved high AUCs and F1-scores, their performance outcomes varied across time intervals. Conclusions: These findings support the feasibility and clinical relevance of using short-term HRV as a noninvasive, data-driven tool for early risk stratification in critical care, potentially guiding timely therapeutic decisions in high-risk ICU patients and thereby reducing in-hospital mortality. Full article

Phycocyanin is a natural, non-toxic, blue pigment-protein with many commercial applications. Its exploitation in various biotechnological sectors strongly depends on its purity grade (P). Phycocyanin is largely used in food industry where a low purity grade is required, while its widespread use in sectors requiring a higher purity is hampered by the cost of large-scale industrial production. Industry, in fact, needs simple, easily scalable and cost-effective procedures to ensure sustainable production of high-quality pigment. In this work we applied the innovative two-step ultrasound-based process UltraBlu to the pilot-scale production of phycocyanin. A total of 50 L of biomass suspension of commercial Spirulina were processed in batch mode. The pigment extract was obtained in one day, including the biomass harvesting. Food/cosmetic grade (P = 1.41–1.76) and a good yield (Y = 59.2–76.1%) were achieved. The initial results obtained suggest that UltraBlu can be an effective scalable process suitable to produce phycocyanin also on an industrial scale. Full article

Poly(ε-caprolactone)-b-polysarcosine (PCL-b-PSar) block copolymers (BCPs) emerge as a promising alternative to conventional poly(ε-caprolactone)-b-poly(ethylene oxide) BCPs for biomedical applications, leveraging superior biocompatibility and biodegradability. In this study, we synthesized two series of PCL-b-PSar BCPs with controlled polymerization degrees (DP of PCL: 45/67; DP of PSar: 28–99) and low polydispersity indexes (Đ ≤ 1.1) and systematically investigated their crystallization-driven self-assembly (CDSA) in alcohol solvents (ethanol, n-butanol, and n-hexanol). It was found that the limited solubility of PSar in alcohols resulted in competition between micellization and crystallization during self-assembly of PCL-b-PSar, and thus coexistence of lamellae and spherical micelles. To overcome this morphological heterogeneity, we developed a modified self-seeding method by employing a two-step crystallization strategy (i.e., T_c1 = 33 °C and T_c2 = 8 °C), achieving conversion of micelles into crystals and yielding uniform self-assembled structures. PCL-b-PSar BCPs with short PSar blocks tended to form well-defined two-dimensional lamellar crystals, while those with long PSar blocks induced formation of hierarchical structures in the PCL₄₅ series and polymer aggregation on crystal surfaces in the PCL₆₇ series. Solvent quality notably influenced the self-assembly pathways of PCL₄₅-b-PSar₂₈. Lamellar crystals were formed in ethanol and n-butanol, but micrometer-scale dendritic aggregates were generated in n-hexanol, primarily due to a significant Hansen solubility parameter mismatch. This study elucidated the CDSA mechanism of PCL-b-PSar in alcohols, enabling precise structural control for biomedical applications. Full article

Automated waste classification is an essential step toward efficient recycling and waste management. Traditional deep learning models, such as convolutional neural networks, rely on extensive labeled datasets to achieve high accuracy. However, the annotation process is labor-intensive and time-consuming, limiting the scalability of these approaches in real-world applications. Zero-shot learning is a machine learning paradigm that enables a model to recognize and classify objects it has never seen during training by leveraging semantic relationships and external knowledge sources. In this study, we investigate the potential of zero-shot learning for waste classification using two vision-language models: OWL-ViT and OpenCLIP. These models can classify waste without direct exposure to labeled examples by leveraging textual prompts. We apply this approach to the TrashNet dataset, which consists of images of municipal solid waste organized into six distinct categories: cardboard, glass, metal, paper, plastic, and trash. Our experimental results yield an average classification accuracy of 76.30% with Open Clip ViT-L/14-336 model, demonstrating the feasibility of zero-shot learning for waste classification while highlighting challenges in prompt sensitivity and class imbalance. Despite lower accuracy than CNN- and ViT-based classification models, zero-shot learning offers scalability and adaptability by enabling the classification of novel waste categories without retraining. This study underscores the potential of zero-shot learning in automated recycling systems, paving the way for more efficient, scalable, and annotation-free waste classification methodologies. Full article

Form-based design codes have emerged as a planning tool aimed at shaping the physical form of neighborhoods to reinforce local character and enhance sense of place (SoP). However, their effectiveness in delivering these outcomes remains underexplored. This study investigates the extent to which Buffalo’s Green Code—a form-based zoning ordinance—enhances SoP in residential environments, using Elmwood Village as a case study. A multi-scalar analytical framework assesses SoP at the building, street, and neighborhood levels. Empirical data were gathered through an online survey, while the neighborhood was systematically mapped into street segment blocks categorized by Green Code zoning. The study consolidates six Green Code classifications into three overarching categories: mixed-use, residential, and single-family. SoP satisfaction is analyzed through a two-step process: first, comparative assessments are conducted across the three zoning groups; second, k-means clustering is applied to spatially map satisfaction levels and evaluate SoP at different scales. Findings indicate that mixed-use areas are most closely associated with place identity, while residential and single-family zones (as defined by the Buffalo Green Code) yield higher satisfaction overall—though satisfaction varies significantly across spatial scales. These results suggest that while form-based codes can strengthen SoP, their impact is uneven, and more scale-sensitive zoning strategies may be needed to optimize their effectiveness in diverse urban contexts. This research overall offers an empirically grounded, multi-scalar assessment of zoning impacts on lived experience—addressing a notable gap in the planning literature regarding how form-based codes perform in established, rather than newly developed, neighborhoods. Full article

Micro-fragmented adipose tissue (MFAT) is a promising autologous therapy for knee osteoarthritis. To avoid repeated liposuction procedures for its clinical application, MFAT obtained from patients with knee osteoarthritis was stored at −80 °C in a tissue bank. This study describes the preparation, cryopreservation, thawing, and washing, as well as comprehensive analysis of cell populations in fresh and MFAT thawed after two years. Immunophenotyping of both fresh and thawed MFAT showed a significant presence of endothelial progenitors and pericytes in the stromal vascular fraction. Viability before (59.75%) and after freezing (55.73%) showed no significant difference. However, the average cell count per gram of MFAT was significantly reduced in thawed samples (3.00 × 10⁵) compared to fresh ones (5.64 × 10⁵), likely due to processing steps. Thawed MFAT samples showed increased CD73 expression on the CD31^highCD34^high subset of EP and SA-ASC, as well as increased expression of CD105 on EP, the CD31^lowCD34^low subset of EP, pericytes, and SA-ASC. Microbiological testing confirmed 100% sterility, and double washing efficiently removed DMSO, confirming sample safety. Histological analysis revealed healthy, uniformly shaped adipocytes with intact membranes. This approach allows accurate estimation of cell yield for intra-articular injection, ensuring delivery of the target cell number into the knee. Quality control analysis confirms that cryopreserved MFAT retains high cellular and structural integrity, supporting its safety and suitability for clinical application. Full article

Mining and analyzing information from multiple sources—such as geophysics and geochemistry—is a key aspect of big data-driven mineral prediction. Clustering, which groups large datasets based on distance metrics, is an essential method in multidimensional data analysis. The Two-Step Clustering (TSC) approach offers advantages by handling both categorical and continuous variables and automatically determining the optimal number of clusters. In this study, we applied the TSC method to mineral prediction in the northeastern margin of the Jiaolai Basin by: (i) converting residual gravity and magnetic anomalies into categorical variables using Ward clustering; and (ii) transforming 13 stream sediment elements into independent continuous variables through factor analysis. The results showed that clustering is sensitive to categorical variables and performs better with fewer categories. When variables share similar distribution characteristics, consistency between geophysical discretization and geochemical boundaries also influences clustering results. In this study, the (3 × 4) and (4 × 4) combinations yielded optimal clustering results. Cluster 3 was identified as a favorable zone for gold deposits due to its moderate gravity, low magnetism, and the enrichment in F1 (Ni–Cu–Zn), F2 (W–Mo–Bi), and F3 (As–Sb), indicating a multi-stage, shallow, hydrothermal mineralization process. This study demonstrates the effectiveness of combining Ward clustering for variable transformation with TSC for the integrated analysis of categorical and numerical data, confirming its value in multi-source data research and its potential for further application. Full article

Seed vigor testing traditionally requires large sample sizes and extended durations. The biospeckle laser (BSL) technique offers a faster, image-based alternative for seed analysis though the standardization of set protocols. This study evaluated the efficiency of local and global BSL analyses in bean seeds (Phaseolus vulgaris L.). Two groups of seeds (872 in total) were classified into high- and low-vigor seeds using the emergence test over 800 samples. The BSL test was then applied to 72 seeds (36 per group), analyzing biological activity locally (vascular and embryonic areas) and globally (whole image). BSL analysis detected significant differences between the groups (p < 0.05). Among the methods, the local analysis of the embryonic axis was most effective (F = 44.252, p = 0.000), showing a clearer distinction than the global analysis (F = 19.484, p = 0.000). The vascular area analysis did not yield significant results. These findings highlight the efficiency of the local BSL analysis at the embryonic axis for vigor tests compared to the global analysis. However, it was observed that the selected point in the local analysis affects the reliability of the vigor test. It was a relevant step toward standardization demanding additional tests in other species and varieties. Full article

Choosing appropriate tillage methods and nitrogen application are important steps in the management of wheat production for obtaining high-yield and high-quality products, as well as managing the level of weed infestation. The aim of this research was to examine the impacts of three different tillage practices (conventional tillage—CT, mulch tillage—MT, and no tillage—NT), and two top dressing fertilization nitrogen levels (rational—60 kg ha⁻¹ and high—120 kg ha⁻¹) on the grain yield and quality of winter wheat, as well as on weed infestation. The present study was carried out in field experiments on chernozem luvic type soil at the Faculty of Agriculture Belgrade-Zemun Experimental field trial “Radmilovac”, in the growing seasons of 2020/2021–2022/2023. The C/N ratio in the soil was also assessed on all plots. The results showed that the number of weeds and their fresh and air-dry weights were higher on the MT and NT plots, compared to the CT plots. Therefore, the CT system has better effects on the yield (5.91 and 5.36 t ha⁻¹) and the protein content (13.3 and 13.1%). Furthermore, the grain weight per spike and the 1000-grain weight were higher in the wheat from the CT system (41.83 and 42.75 g) than from the MT (40.34 and 41.49 g) and NT (40.26 and 41.08 g) systems. Also, the crops from the CT system had higher values of grain density and grain uniformity compared to the crop from the MT and NT systems. Fertilization with a high nitrogen level (120 kg ha⁻¹) causes higher grain yield and more weediness compared with the rational level (60 kg ha⁻¹). Top dressing fertilization in each tillage system resulted in an increase in the number of weeds, but, at the same time, it also resulted in stronger competitive ability of the wheat crop against weeds. The most favorable C/N ratio occurred on the NT plots, and the least beneficial one on the CT ones. A correlation analysis showed strong negative correlations of number (r = −0.82) and fresh weed mass (r = −0.72) with yield. It is concluded that the conventional tillage practice with a low nitrogen dose manifests its superior performance in minimizing weed infestation and maximizing crop productivity. Full article

Understanding the impacts of the spatial non-stationarity of environmental factors on surface soil moisture (SSM) in different seasons is crucial for effective environmental management. Yet, our knowledge of this phenomenon remains limited. This study introduces an interpretable machine learning framework that combines the SHapley Additive exPlanations (SHAP) method with two-step clustering to unravel the spatial drivers of SSM across Iran. Due to the limited availability of in situ SSM data, the performance of three global SSM datasets—SMAP, MERRA-2, and CFSv2—from 2015 to 2023 was evaluated using agrometeorological stations. SMAP outperformed the others, showing the highest median correlation and the lowest Root Mean Square Error (RMSE). Using SMAP, we estimated SSM across 609 catchments employing the Random Forest (RF) algorithm. The RF model yielded R² values of 0.89, 0.83, 0.70, and 0.75 for winter, spring, summer, and autumn, respectively, with corresponding RMSE values of 0.076, 0.081, 0.098, and 0.061 m³/m³. SHAP analysis revealed that climatic factors primarily drive SSM in winter and autumn, while vegetation and soil characteristics are more influential in spring and summer. The clustering results showed that Iran’s catchments can be grouped into five categories based on the SHAP method coefficients, highlighting regional differences in SSM controls. Full article

Detecting leakage using a point cloud acquired by mobile laser scanning (MLS) presents significant challenges, particularly from within three-dimensional space. These challenges primarily arise from the prevalence of noise in tunnel point clouds and the difficulty in accurately capturing the three-dimensional morphological characteristics of leakage patterns. To address these limitations, this study proposes a classification method based on XGBoost classifier, integrating both intensity and geometric features. The proposed methodology comprises the following steps: First, a RANSAC algorithm is employed to filter out noise from tunnel objects, such as facilities, tracks, and bolt holes, which exhibit intensity values similar to leakage. Next, intensity features are extracted to facilitate the initial separation of leakage regions from the tunnel lining. Subsequently, geometric features derived from the k neighborhood are incorporated to complement the intensity features, enabling more effective segmentation of leakage from the lining structures. The optimal neighborhood scale is determined by selecting the scale that yields the highest F_1-score for leakage across various multiple evaluated scales. Finally, the XGBoost classifier is applied to the binary classification to distinguish leakage from tunnel lining. Experimental results demonstrate that the integration of geometric features significantly enhances leakage detection accuracy, achieving an F_1-score of 91.18% and 97.84% on two evaluated datasets, respectively. The consistent performance across four heterogeneous datasets indicates the robust generalization capability of the proposed methodology. Comparative analysis further shows that XGBoost outperforms other classifiers, such as Random Forest, AdaBoost, LightGBM, and CatBoost, in terms of balance of accuracy and computational efficiency. Moreover, compared to deep learning models, including PointNet, PointNet++, and DGCNN, the proposed method demonstrates superior performance in both detection accuracy and computational efficiency. Full article

Two-dimensional (2D) materials have been proposed for use in a multitude of applications, with graphene being one of the most well-known 2D materials. Despite their potential to contribute to innovative solutions, the fabrication of such devices still faces significant challenges. One of the key challenges is the fabrication at a wafer-level scale, a fundamental step for allowing reliable and reproducible fabrication of a large volume of devices with predictable properties. Overcoming this barrier will allow further integration with sensors and actuators, as well as enabling the fabrication of complex circuits based on 2D materials. This work presents the fabrication steps for a process that allows the on-wafer fabrication of active and passive radiofrequency (RF) devices enabled by graphene. Two fabrication processes are presented. In the first one, graphene is transferred to a back gate surface using critical point drying to prevent cracks in the graphene. In the second process, graphene is transferred to a flat surface planarized by ion milling, with the gate being buried beneath the graphene. The fabrication employs a damascene-like process, ensuring a flat surface that preserves the graphene lattice. RF transistors, passive RF components, and antennas designed for backscatter applications are fabricated and measured, illustrating the versatility and potential of the proposed method for 2D material-based RF devices. The integration of graphene on devices is also demonstrated in an antenna. This aimed to demonstrate that graphene can also be used as a passive device. Through this device, it is possible to measure different backscatter responses according to the applied graphene gating voltage, demonstrating the possibility of wireless sensor development. With the proposed fabrication processes, a flat graphene with good quality is achieved, leading to the fabrication of RF active devices (graphene transistors) with intrinsic f_T and f_max of 14 GHz and 80 GHz, respectively. Excellent yield and reproducibility are achieved through these methods. Furthermore, since the graphene membranes are grown by Chemical Vapor Deposition (CVD), it is expected that this process can also be applied to other 2D materials. Full article

The ability to study mature neuronal cells ex vivo is complicated by their non-dividing nature and difficulty in obtaining large numbers of primary cells from organisms. Thus, numerous transformed progenitor models have been developed that can be routinely cultured, then scaled, and differentiated to mature neurons. In this paper, we present a new method for differentiating one such model, the Lund human mesencephalic (LUHMES) dopaminergic neurons. This method is two days faster than some established protocols, results in nearly five times greater numbers of mature neurons, and involves fewer handling steps that could introduce technical variability. Moreover, it overcomes the problem of cell aggregate formation that commonly impedes high-resolution imaging, cell dissociation, and downstream analysis. While recently established for herpes simplex virus type 1, we demonstrate that LUHMES neurons can facilitate studies of other herpesviruses, as well as RNA viruses associated with childhood encephalitis and hemorrhagic fever. This protocol provides an improvement in the generation of large-scale neuronal cultures, which may be readily applicable to other neuronal 2D cell culture models and provides a system for studying neurotrophic viruses. We named this method the Streamlined Protocol for Enhanced Expansion and Differentiation Yield, or SPEEDY, method. Full article