Search Results (59,311)

Background: Peer-assisted learning (PAL) is used in medical education. Its impact in basic medical sciences remains unexplored, namely considering the perspectives of all the populations involved: students, undergraduate teaching assistants (UTAs), and faculty members. We evaluated the educational impact of a PAL programme in Histology and Embryology in a medical course at the Faculty of Medicine of the University of Porto (FMUP). Methods: A cross-sectional convergent mixed-methods study was conducted in the Histology and Embryology course. Tailored online questionnaires comprising Likert-type items and open-ended questions were replied to by students attending theoretical–practical classes with UTAs, the UTAs, and the professors. Internal consistency was assessed using Cronbach’s alpha. Students’ theoretical–practical examination scores were compared between classes with and without UTAs using the Mann–Whitney U test, and effect size (r) was calculated to estimate the magnitude of differences observed. Results: Students (n = 190) reported highly positive perceptions regarding the creation of a more approachable learning environment, clarification of doubts, and identification of histological structures and strongly recommended the extension of UTAs to other courses of the medical school. UTAs (n = 17) described gains in disciplinary understanding, broader perspectives on the medical curriculum, communication and public-speaking skills, teamwork, leadership, self-confidence, and interest in academic careers. Professors (n = 6) valued PAL for improving individual support, facilitating time management, and contributing to UTA training, while highlighting the need for structured pedagogical preparation. Students attending classes with UTAs achieved significantly higher theoretical–practical examination scores (p = 0.04; effect size r = 0.12). Conclusions: PAL was perceived as highly beneficial by all groups involved in the project, enhancing the learning environment, supporting knowledge consolidation, and developing pedagogical and interpersonal skills. A grade analysis indicated that PAL was associated with improved academic performance. These findings reinforce the value of integrating PAL initiatives into preclinical medical education while highlighting the importance of sustained tutor preparation and supervision. Full article

This study compares gold-backed and conventional cryptocurrencies in terms of dependence structures and hedging effectiveness relative to precious metals. Daily data for gold, silver, cryptocurrencies, gold-backed cryptocurrencies, and USD-backed stablecoins from July 2020 to March 2026 are analyzed using a multivariate stochastic volatility framework with a grouped factor structure. Gold-backed cryptocurrencies move closely with gold and silver and provide meaningful hedging benefits. Conventional cryptocurrencies present weaker and less stable relationships with precious metals, reducing hedging potential. Important differences emerge between gold and silver, suggesting that precious metals should not be treated as a homogeneous asset class. Gold-backed cryptocurrencies appear much more closely aligned with precious metals than conventional cryptocurrencies. Additional analyses show that hedging effectiveness increases substantially during periods of elevated volatility, particularly for PAXG and XAUT, indicating stronger risk-reduction benefits under stressed market conditions. Robustness tests using SPDR Gold Shares (GLD) confirm the stability of the main findings. The findings are relevant for portfolio diversification, hedging decisions, and risk management. Full article

DeFi perpetual futures have expanded from crypto-native instruments to tokenised equities and commodities, yet the economics of these instruments remain poorly understood. We study 17 assets—5 crypto coins, 8 tokenised equities, and 4 tokenised commodities—on three DeFi perpetual platforms (Hyperliquid, EdgeX, Lighter) over July 2025 to February 2026. Applying a rolling 3-day t-test to identify abnormal trading volume without a predetermined event calendar, we document 1797 statistically significant volume anomalies. DeFi perpetual volume is driven primarily by macroeconomic and policy shocks (ADA $t=+628$ on the U.S. Crypto Strategic Reserve announcement; 15 of 17 assets simultaneously anomalous during January 2026 mega-cap earnings), asset-class-specific catalysts, and a recurring 24/7 market-structure effect tied to weekends and U.S. holidays. Price tracking accuracy reveals a sharp maturity gradient: crypto coin perpetuals exhibit near-perfect price tracking ($\rho \ge 0.999$) and strong TradFi volume co-movement ($\rho \left(0\right)\in [0.72,0.83]$), while equity perpetuals show weaker integration and commodity perpetuals range from adequate (oil, gold) to unreliable (natural gas). We conclude that crypto DeFi perpetuals constitute credible synthetic economic claims on underlying assets, while equity and commodity perpetuals remain at an early developmental stage. Integration with traditional financial markets is well-established for crypto coin perpetuals; for equity and commodity perpetuals, the evidence is preliminary, given short observation windows, and further research with longer time series is needed before definitive conclusions can be drawn. Full article

Bridge health monitoring and machine learning are increasingly intertwined for civil engineers and artificial intelligence experts. Bridges’ poor health can result in severe outcomes if not addressed in time. Therefore, continuous monitoring is required to detect any anomaly or damage. Sensors, such as accelerometers, inclinometers, thermistors, etc., can help actively monitor these bridges. The signals from these sensors help record physiological activities. Such activities are helpful for anomaly detection, damage localization, and bridge health predictions with the help of machine learning algorithms. The proposed method extracts features from the dynamic response of a bridge to ambient excitation. It focuses on processing the signal received from different accelerometers installed on a steel railway bridge to determine the location of the damage and the level of the damage predictions. Initially, features are extracted from time-series data; then, they are fed to a deep neural network after some pre-processing. Normal and augmented data are used with different parameter tuning for results. Original data is also subdivided, and the effect of data slicing on the predictions is investigated. The results show that one-fourth of the slicing of the original data gives the best results for training and testing accuracy with a deep neural network. The results show that the reduced matrix representation, particularly the 40 × 40 feature slicing, improved the classification performance for the predefined bridge scenario classes under the considered experimental settings. For bridge scenario classification, the best reported accuracy was 93.54%, while for damage intensity classification the best reported accuracy was 98.21%. In the DNN-based optimizer comparison, the Adam optimizer achieved higher and more stable performance than Stochastic Gradient Descent (SGD), with test accuracies of 92.3% and 93.7% compared with 75.2% and 86.4%, respectively. It is also observed that the Adam optimizer outperformed Stochastic Gradient Descent (SGD) in terms of both damage localization and damage intensity estimation. Full article

Few-shot remote sensing scene classification aims to accurately recognize unseen scene categories using only a scarce number of labeled samples, which has emerged as a research hotspot in the field of remote sensing image interpretation. However, remote sensing images intrinsically suffer from large intra-class variations, high inter-class similarities, and complex background interferences. Traditional few-shot learning methods typically perform feature metric learning in Euclidean space, making it difficult to capture the non-Euclidean geometric distribution characteristics of remote sensing features, and they often neglect the spatial structural information embedded in feature maps. To address these issues, this paper proposes a novel few-shot remote sensing scene classification method, termed ZSFS-RGBN, which integrates a Zigzag Scanning Feature Sequence with a Riemannian Geometric Barycenter Network. Specifically, ResNet12 is first employed as the backbone to extract deep convolutional feature maps from both the support and query sets. Second, a Zigzag scanning strategy is introduced to reorganize the two-dimensional feature maps into one-dimensional feature sequences, thereby effectively preserving the spatial locality and structural continuity of the features. Third, an autoregressive moving average (ARMA) model is constructed to characterize the spatial dependencies of the feature sequences, and its state parameters are mapped onto a symmetric positive definite (SPD) matrix manifold, enabling the deep semantic representations of remote sensing scenes in a non-Euclidean geometric space. Finally, a Riemannian geometric barycenter network is designed to learn the Riemannian barycenter of each category on the SPD manifold, where a joint loss function is introduced to simultaneously optimize intra-class compactness and inter-class separability. Comprehensive experiments are conducted on three public remote sensing scene datasets: NWPU-RESISC45, UC Merced Land-Use, and WHU-RS19. Experimental results demonstrate that the proposed method consistently outperforms several representative state-of-the-art approaches under both 5-way 1-shot and 5-way 5-shot settings. Furthermore, ablation studies verify the effectiveness of each component within the proposed framework. Full article

Automatic respiratory-sound classification becomes particularly challenging when crackles and wheezes coexist within the same respiratory cycle, because overlap events contain mixed acoustic characteristics and are severely underrepresented in standard benchmarks. In the ICBHI 2017 Respiratory Sound Database, the rare overlap label, Both, is frequently absorbed into Crackle, Wheeze, or Normal classes by conventional flat classifiers, resulting in poor interpretability and weak clinical reliability. To address this issue, this paper proposes OCAD (Overlap Composition and Class-Atom Decoding), an explainable overlap-aware respiratory-sound classification framework. Unlike conventional end-to-end classifiers that directly map features to labels, OCAD explicitly decomposes respiratory-sounds into interpretable crackle and wheeze factors, composes an overlap representation from these abnormal-event components, and performs classification through structured class atoms whose geometric relationships reflect the progression from Normal to single-event and overlap states. By preserving a fixed acoustic backbone across all compared systems, the study isolates the contribution of the proposed explainable representation and decoding mechanism. Experimental results on a fixed patient-level split of the ICBHI 2017 dataset show that OCAD achieves an ICBHI Score of 0.6920, Macro Score of 0.6550, Macro F1 of 0.6680, crackle sensitivity of 0.6300, wheeze sensitivity of 0.6550, and Both sensitivity of 0.2620. Compared with the strongest structured class-atom baseline under the same protocol, OCAD improves the ICBHI Score by 0.1334, Macro F1 by 0.2575, and Both sensitivity by 0.2054 absolute points. Additional robustness analyses across repeated seeds, patient-level cross-validation, and external respiratory-sound datasets further support the effectiveness of the proposed framework. The proposed overlap-aware representation and decoding strategy provides a more interpretable and reliable approach for respiratory sound classification in the presence of co-occurring abnormal events. Full article

We revise the Soft Landing Problem deriving the Optimal Guidance Policy and the Value Function of the minimum time problem in its Controllable Set. Using this Guidance Policy the lander can be guided not only along the nominal trajectory or in its neighborhood but in all conditions that can have a possible successful landing, thus providing a safer approach to changes in planning, failures or other unknown conditions. In addition to the classic solutions of this problem we have fully described a new class of broken extremals which occur when the lander is falling with an high vertical speed and can be used in case of failures or unplanned conditions. Full article

Background: Integrating human neural signals with computational vision systems offers a promising route toward more robust visual recognition, yet supporting mixed-granularity recognition, where both coarse- and fine-grained categories must be distinguished within a unified system, remains challenging due to the heterogeneous feature spaces of electroencephalography (EEG) and visual data. Methods: We propose “Align and Fuse,” a two-stage Transformer-based framework. Stage 1 constructs a shared semantic space using a hardness-aware multimodal supervised contrastive loss with Hard Negative Weighting to explicitly target confusable class pairs. Stage 2 employs a multimodal Transformer with co-attention to fuse the aligned features for classification. Results: On the 80-class EEG-ImageNet benchmark, our framework achieved 91.12% Top-1 accuracy under a temporally separated control protocol, improving over the corresponding vision-only (89.08%) and Standard Transformer (89.95%) baselines. Under the original stratified random split, it achieved 92.56% Top-1 accuracy; on the 40-class EEGCVPR dataset, accuracy reaches 95.82%. Cross-subject experiments yield 90.92% average Top-1 accuracy on four unseen subjects, and Grad-CAM analysis suggests that aligned EEG signals shift the model’s attention toward semantically relevant regions. Conclusions: Coupling hardness-aware alignment with decoupled multimodal fusion supports EEG-augmented recognition by leveraging complementary stimulus-related information under the evaluated protocols. Because EEG features are required at inference time, the framework is positioned as a human-in-the-loop EEG-augmented recognition system rather than a standalone vision model. Full article

As a hundred-kilowatt-class advanced electric propulsion technology, the Variable Specific Impulse Magnetoplasma Rocket (VASIMR) holds immense potential for deep space exploration. During the ion cyclotron resonance heating (ICRH) process of VASIMR, the resonance of high-mass, low-charge-state ions demands exceptionally strong background magnetic fields; its reliance on strong magnetic fields imposes stringent thermal control requirements on superconducting magnets, significantly driving up the volume and mass penalties of the system. To address this challenge, this study explores a low-magnetic-field VASIMR architecture utilizing a light gas, Neon (Ne), as the propellant. We systematically investigate the thrust performance and the evolution of ion energy under the multivariable coupling of pre-ionization power, ICRH power, and the magnetic field topology within the resonance zone. The results demonstrate the technical and engineering feasibility of the low-magnetic-field and light-propellant propulsion scheme. Specifically, the thrust gains corresponding to the pre-ionization and ICRH power are approximately 2.2 mN/100 W and 10.3 mN/500 W, respectively. Furthermore, optimizing the magnetic field topology significantly enhances the ion energy absorption efficiency in the resonance zone, yielding a thrust improvement of 25.3 mN. This study achieves a significant reduction in the background magnetic field strength compared to conventional VASIMR, elucidating the multi-regime control mechanisms of the low-field VASIMR. These findings lay a robust theoretical and experimental basis for future lightweight designs and performance leaps. Full article

The objective of this study is to recover the transient out-of-plane displacement field of clamped circular plates subjected to blast loading using a single high-speed camera, as a low-cost alternative to stereo Digital Image Correlation (DIC) for the specific class of axisymmetrical structural responses of circular plates. The dynamic response of thin metal plates to blast loading is a fundamental problem in protective structural design, traditionally investigated through DIC. Although it provides full-field displacement measurements with high spatial resolution, it requires stereo camera arrangements, controlled illumination, speckle pattern preparation, and elaborate calibration procedures that significantly increase experimental cost and complexity. This study introduces a monocular optical method applicable to axisymmetrically defined material testing applications, such as the response of circularly supported isotropic plates under a uniform impulsive load, to recover the transient out-of-plane displacement field without using DIC. Clamped circular aluminum plates are subjected to blast loading generated by PG-3 charges of variable mass detonated at the closed end of a shock tube, with the exposed face matching the tube cross-section so as to enforce axisymmetric pressure load. A diametral reference line marked on the rear face of each specimen was recorded by a single high-speed camera, and a perspective correction derived from the axisymmetric deformed geometry was then applied to reconstruct the time-resolved displacement profile along the diameter. The permanent post-test deformed shape of each plate was subsequently digitized through 3D scanning and used as ground truth to validate the optical reconstruction. The reconstructed profiles closely matched the scans: for the conventional responses the root-mean-square error was 1.251 mm with a normalized mean residual of 6.57% (Case A) and 1.793 mm (9.20%, Case B), while for the anomalous counterintuitive response it was 1.043 mm (14.93%, Case C). Symmetry can thus be exploited as an active measurement principle to obtain quantitative blast-response data with substantially reduced experimental burden and without specialized stereo-optical instrumentation. Full article

Background: Amiodarone hydrochloride is an antiarrhythmic agent, primarily classified as a Vaughan Williams Class III antiarrhythmic agent, used for the treatment and prevention of arrhythmia and designated as an essential national drug. Recently, the need to develop formulations suitable for continuous manufacturing, which is gaining significant attention in the pharmaceutical industry, has emerged. Objective: The objective of this study was to evaluate the formulation potential of amiodarone hydrochloride for the development of an oral tablet, with a specific focus on improving the initial dissolution rate to design a formulation optimized for continuous manufacturing. Methods: The primary physicochemical properties of amiodarone hydrochloride were predicted and subsequently validated through experimental characterization. Furthermore, the ratios of the binder and granulation liquid were optimized to facilitate robust and successful production via continuous manufacturing. Results: A new amiodarone hydrochloride tablet formulation, optimized for a continuous wet granulation process, was successfully developed by improving the initial dissolution rate through the optimization of the respective amounts of binder and granulation liquid. Conclusions: The optimal formulation design established in this research provides a critical foundation for enhancing the applicability of a continuous process in the manufacturing of amiodarone hydrochloride tablets. Full article

Brain tumor classification from magnetic resonance imaging (MRI) remains challenging in settings where only image-level labels are available and tumor classes exhibit overlapping visual characteristics. In this study, we consider the publicly available Brain Tumor MRI Dataset from Kaggle, a four-class dataset composed of 2D MRI slices belonging to the categories glioma, meningioma, pituitary tumor, and no tumor. Accordingly, the proposed framework is formulated as a slice-based multiclass classification approach rather than a volumetric 3D analysis pipeline. We propose a lightweight and interpretable framework that integrates handcrafted multiscale MRI descriptors, an artificial neural network (ANN), Bee Colony Optimization (BCO)-based neural architecture search, and fuzzy softmax confidence modeling. Each MRI slice is represented by a compact 9-dimensional feature vector derived from intensity, local entropy, and gradient magnitude computed globally and over non-overlapping spatial blocks. The ANN design problem is formulated as a discrete–continuous optimization task, where BCO is employed to optimize network architecture and training hyperparameters by maximizing validation macro-$F_1$. To quantify predictive reliability, the softmax outputs are interpreted as fuzzy class memberships and further analyzed using maximum membership, normalized entropy, decision margin, and ambiguity measures, enabling confidence-aware reliability assessment. These fuzzy confidence descriptors enable confidence-threshold-based selective classification and rejection of low-confidence predictions. Across repeated runs, the optimized BCO-ANN achieved a mean test accuracy of $0.781\pm 0.009$, mean macro-$F_1$ of $0.775\pm 0.010$, mean Brier score of $0.319\pm 0.012$, and mean Expected Calibration Error (ECE) of $0.0273\pm 0.0080$, compared with $0.748\pm 0.011$, $0.738\pm 0.013$, $0.352\pm 0.010$, and $0.0446\pm 0.0071$ for the baseline ANN, respectively. Under confidence-threshold-based rejection, selective macro-$F_1$ increased to $0.820\pm 0.009$ at $\tau =0.55$ and to $0.874\pm 0.020$ at $\tau =0.85$, with the expected reduction in coverage. These results indicate that the proposed framework provides a transparent and reproducible approach for optimization-aware and confidence-aware multiclass brain tumor MRI classification in a lightweight handcrafted feature setting. Full article

Adaptive and adverse brain states are often assumed to lie on a shared molecular continuum, but this assumption has rarely been evaluated against explicit transcriptomic alternatives. This study aimed to compare two representations of cross-context brain transcriptomic organization: a transcriptome-wide global-axis model and a low-dimensional reciprocal model. We benchmarked these models across a curated cross-study brain cohort spanning exercise, alcohol-related adversity-like contexts, stress, aging, and neurodegeneration, using prespecified intervention-like and adversity-like directional contrast labels rather than assuming homogeneous biological states. We assessed the competing representations using signed-effect correlations, permutation analyses, non-linear fitting, and held-out reconstruction, and we then examined the resulting structure through region-specific human bulk evaluation and exploratory cellular, single-nucleus, spatial, and chromatin projection analyses. These downstream analyses were used to examine localization and biological interpretability and were not treated as independent evaluation of the module 1/module 2 (M1/M2) partition. The combined signed-effect statistics were interpreted as representation-level directional summaries rather than estimates of a homogeneous cross-study biological effect. The global-axis model received limited support: intervention-like and adversity-like signed-effect summaries were only weakly correlated, were not stronger than permutation null expectations, and were not improved by non-linear fitting. Within the selected reciprocal-gene space, a rank-1 latent profile reconstructed held-out genes more accurately than the hard M1/M2 partition, whereas the M1/M2 discretization provided a more interpretable but selection-conditioned directional summary. Human analyses yielded an asymmetric pattern: a significant M1 association was observed only in the hippocampal dataset, whereas M2, the reciprocal index, and the other examined brain regions showed no consistent corresponding effects; leave-one-stratum-out analyses indicated poor cross-stratum reproducibility of the exact gene-level partition. These findings motivate a low-dimensional reciprocal representation as an exploratory framework while emphasizing context dependence, cohort dependence, and heterogeneity. Full article

PFAS form a class of synthetic chemicals that has become an area of increasing concern because of its impact on the environment and the threat it poses to human health. The structure of PFAS makes them highly resistant to degradation. As a result, they are highly effective at bioaccumulation. Certain water treatment technologies have been proven to remove PFAS from contaminated water sources. This study reviews the most promising treatment technologies used for the treatment of PFAS-contaminated waters. Well-established treatment technologies, such as granular activated carbon, ion exchange resin, reverse osmosis, and nanofiltration, were quantitatively compared. The removal efficiency was assessed by collecting the data of individual PFAS species from the literature and grouping them into five groups: PFAS (all species), PFSA, PFCA, long chain, and short chain. The results identified that, for all PFAS groups, the most effective treatment technologies were in the following order: reverse osmosis, nanofiltration, ion exchange resin, and granular activated carbon. The performance of reverse osmosis and nanofiltration did not appear to significantly differ between the different PFAS groups, as opposed to ion exchange resin and granular activated carbon, where there was a greater degree of variation in performance between different PFAS groups. Overall, it was identified that membrane technologies outperformed adsorbent technologies. However, the cost associated with membrane technologies may limit its economic viability when compared with adsorbent technologies, which are typically a more viable option except under specific circumstances. For example, contaminated water with high concentrations of other contaminants that need to be treated simultaneously. Lack of standardised experimental and operational conditions limited the available data. While this work provides guidance on which treatment is more likely to be appropriate based on the concentration and composition of different species of PFAS, more data are needed to conduct a more accurate statistical analysis and enable accurate modelling of treatment performance. Full article

The escalation of antimicrobial resistance (AMR) represents a serious global threat to public health, with AMR-associated mortality estimated to increase by 70% by 2050. As pathogens evolve through enzymatic inactivation, target modification, efflux-mediated clearance, biofilm formation, and broader genetic adaptation, conventional therapies are increasingly compromised, while the antibiotic development pipeline remains critically constrained by high discovery and development costs, weak commercial incentives, and the escalating complexity of resistance mechanisms. This review comprehensively synthesizes advanced pharmacological and biotechnological innovations designed to circumvent these entrenched resistance mechanisms. We highlight the development of novel therapeutic classes, particularly lariocidin, which disrupts bacterial protein synthesis via a previously unexploited ribosomal-binding site. Moreover, we critically evaluate molecular interventions, emphasizing CRISPR/Cas-based gene silencing and genome editing as precise tools to neutralize specific resistance determinants, such as the mecA gene in methicillin-resistant Staphylococcus aureus (MRSA). Concurrently, we explore the integration of engineered nanoparticles to revitalize existing antimicrobials by overcoming biofilm barriers, improving drug solubility, and enabling targeted delivery. Collectively, mastering the evolving AMR landscape requires a multidimensional framework that seamlessly integrates these novel molecular targets with advanced rapid diagnostics and robust international governance. Full article