Abstract Sustainable development demands research into safe, renewable energy sources. Wood briquettes offer numerous advantages, but they can contain heavy metal(oid)s, posing environmental challenges, particularly in the ash produced during combustion. This study examines the concentrations of heavy metal(oid)s (Cd, Cr, Cu, Fe, Mn, Ni, Pb, Co, Zn, and As) in wood briquettes and their residual ash. Samples were prepared via wet digestion using 65% nitric acid (HNO3) in polytetrafluoroethylene vessels, followed by analysis using flame and graphite furnace atomic absorption spectrometry. The results showed that arsenic (As) had the lowest concentration in wood briquettes, while iron (Fe) was the highest. In the ash, chromium (Cr) was detected at the lowest concentration (0.80 mg/kg), while iron (Fe) reached 5830 mg/kg. Heavy metal concentrations in wood briquettes often exceeded permissible limits, and the concentrations in ash were significantly higher, making some ash samples unsuitable for agricultural use. The ash content ranged from 0.70% to 2.34%. This study provides valuable quantitative data on heavy metal(oid)s before and after combustion, highlighting their potential environmental impact and emphasizing the need for careful management of wood briquette ash.

Psychological interventions represent a core component of contemporary interdisciplinary chronic pain treatment, yet treatment initiation following referral to pain psychology services remains consistently low. Empirical studies across behavioral health and pain medicine demonstrate that referral alone is insufficient to ensure patient engagement with psychological care. This gap between referral and treatment initiation represents a major implementation barrier limiting the impact of evidence-based psychological pain interventions. The present article synthesizes contemporary literature on behavioral health treatment initiation and chronic pain psychology to propose a structured engagement framework designed to improve initiation rates following referral. Using a targeted narrative review methodology, empirical literature published between 2021 and 2025 was examined to identify key determinants of treatment initiation across pain medicine and integrated behavioral health settings. Findings indicate that treatment initiation is best conceptualized as a multistep process involving referral communication, structural and attitudinal barriers, patient readiness, psychoeducation, and system-level facilitation. Evidence from collaborative care models suggests that active engagement strategies embedded within medical workflows can substantially improve treatment initiation rates compared with passive referral approaches. The proposed Active Engagement Model of Pain Psychology Referral integrates individual-level and system-level interventions designed to address common barriers to treatment initiation. Improving initiation requires a shift from passive referral models toward proactive engagement strategies embedded within interdisciplinary pain care. Implementing structured engagement approaches may substantially improve access to evidence-based psychological interventions for chronic pain.

Data scarcity and weak supervision continue to limit the performance of machine learning models in many real-world applications, such as mammography, where Multiple Instance Learning (MIL) often offers the best formulation. While recent foundation models provide strong semantic representations out of the box, effective augmentation of such representations of MIL data remains limited, as existing methods operate at the instance level and fail to capture intra-bag dependencies. In this work, we introduce SetFlow, a generative architecture that models entire MIL bags (i.e., sets) directly in the representation space. Our approach leverages the flow matching paradigm combined with a Set Transformer-inspired design, enabling it to handle permutation-invariant inputs while capturing interactions between instances within each bag. The model is conditioned on both class labels and input scale, allowing it to generate coherent and semantically consistent sets of representations. We evaluate SetFlow on a large-scale mammography benchmark using a state-of-the-art MIL-PF classification pipeline. The generated samples are shown to closely match the original data distribution and even improve downstream performance when used for augmentation. Furthermore, training on synthetic data alone shows competitive results, demonstrating the effectiveness of representation-space generative modeling for data-scarce and privacy-sensitive tasks.

The growing global demand for effective and safe therapeutics has accelerated advances in biomaterials for drug delivery applications. Biomaterials, including polymers, metals, ceramics, and composites, play a central role in modern medical devices and therapeutic systems by enabling controlled interactions with biological environments. Initially defined as inert materials interfacing with biological systems, biomaterials are now rationally engineered to treat, replace, or evaluate tissue and organ functions. Recent progress in regenerative medicine, nanotechnology, and precision healthcare has expanded their use in drug delivery, where tunable physicochemical properties—such as degradation kinetics, surface chemistry, and mechanical stability—allow controlled release, protection of labile therapeutics, and enhanced accumulation at target sites. Polymer-based biomaterials enable sustained drug release through diffusion-controlled, degradation-mediated, or stimulus-responsive mechanisms, thereby extending therapeutic exposure and reducing systemic dosing frequency compared with conventional formulations. Nanostructured carriers, including liposomes, micelles, and dendrimers, further enhance drug delivery by improving solubility, cellular uptake, and site-specific targeting via size control, surface functionalization, and ligand-mediated interactions. Despite these advances, clinical translation remains limited by challenges related to immune–biomaterial interactions, batch-to-batch variability, long-term biodegradation behavior, and the scalability of manufacturing under regulatory constraints. Future biomaterial development must therefore emphasize precision fabrication, good manufacturing practice–compatible production, and biologically informed design strategies that account for patient-specific variability. This review provides a focused overview of biomaterial-based drug delivery systems, summarizes recent technological advances, and critically discusses mechanistic and translational challenges, including immune compatibility, degradation control, and regulatory compliance, with particular emphasis on their implications for personalized drug delivery.

Context. Gamma-ray burst GRB 241030A ( z  = 1.411) exhibited a particularly bright afterglow (similar to the ‘BOAT’, GRB 221009A), detected across gamma-ray, X-ray, UV, and optical bands. The extensive, multi-wavelength observations of this remarkable event provide a valuable opportunity to advance our understanding of GRB afterglow physics. Aims. We aim to constrain the physical properties of the jet, its microphysics, and the characteristics of the circumburst environment in the context of forward-shock emission. Methods. We compiled multi-wavelength observations spanning from a minute to a week after the prompt emission, processing the data through a unified photometry pipeline. Leveraging this comprehensive dataset, we analysed the observations both analytically and using Bayesian inference with two independent models. Our models assume that the afterglow emission arises from the strong forward shock of a laterally structured jet, with possible contributions from synchrotron self-Compton (SSC) scatterings. Results. We find that our models do reproduce the afterglow observations accurately, from the X-rays to the optical, favouring a jet propagating into a constant-density interstellar medium, with a viewing angle within the jet core. However, both analyses – with and without the inclusion of SSC scatterings – require parameter values that are extreme compared to expectations from standard theory. In particular, our results imply extremely energetic jets despite regular prompt energy, leading to a very inefficient prompt emission. Furthermore, the jets are particularly inefficient at accelerating particles, with low ϵ e and ϵ B , leading to significant SSC emission. Finally, our analyses indicate that the jets have large opening angles and propagate in high-density media. Conclusions. If the afterglow is indeed powered by radiation emitted behind a strong forward shock, our results place GRB 241030A within a sub-class of GRBs characterised by extreme kinetic energies, large jet opening angles, and very low prompt emission efficiencies, below 10 −3 , with strong SSC radiation. These predictions are difficult to reconcile with typical expectations from other GRBs. We therefore suggest that the afterglow of GRB 241030A is not solely powered by forward shock emission, and we discuss other options such as a long-lasting reverse-shock contribution.

Perfluorination of the terminal methyl group in ethanol gives rise to different thermodynamics of mixing with water. To understand its origin, we probe structure, thermodynamics and collective vibration modes in aqueous solutions of ethanol (EtOH) or 2,2,2-trifluoroethanol (TFE) by using Terahertz (THz) spectroscopy and molecular dynamics simulations. The THz spectra show mainly two features: a mostly entropy-related peak (below 200 cm-1) related to weaker water-water hydrogen bonds, and an enthalpy-related large band above 200 cm-1 related to water-solute hydrogen bonds. The entropic feature is red-shifted for TFE relative to EtOH, consistent with TFE's subpopulation of weaker solvation shell water-water hydrogen bonds found in the simulations. By contrast, the thermodynamics of mixing is dominated by three effects: the higher probability of forming water-water hydrogen bonds in the solvation shell of either solute than in the bulk; the fact that TFE induces a smaller perturbation per water molecule than EtOH, despite perturbing a slightly larger number of water molecules than EtOH, and TFE's weaker solute-water hydrogen bond. The three effects determine the more negative (favourable) enthalpy of mixing and more negative (unfavourable) entropy of mixing of EtOH relative to TFE at low concentrations. The results confirm that hydrophobic solvation of perfluorinated groups is fundamentally different from that of their alkylated equivalents and have implications for the development of models to predict solubility of perfluorinated molecules.

Malware detection using deep learning faces challenges in model selection for practical deployment. We systematically compare five transfer learning architectures (VGG16, ResNet50, DenseNet121, MobileNetV2, EfficientNetB0) on the MaleBin RGB malware dataset ($\text{1 2, 0 0 0 +}$ images through March 2025). Experiments on NVIDIA A100 GPU evaluated accuracy, efficiency, and deployment suitability. DenseNet121 achieved highest accuracy ($91.20 \%, 8 \mathrm{M}$ parameters), MobileNetV2 provided optimal edge deployment (90.39 %, 3.5 M parameters), while ResNet50 and EfficientNetB0 unexpectedly underperformed $(77.34 \%, 71.16 \%)$. Directions for practitioners are to deploy DenseNet121 for cloud environments, prioritizing accuracy, and MobileNetV2 for resource-constrained edge devices.

Deep learning has become increasingly significant in clinical medicine, including breast cancer detection, offering significant potential to improve patient outcomes. However, recurrent architectures like LSTM (Long Short-Term Memory) and BiLSTM (Bidirectional Long Short-Term Memory) remain underutilized for breast cancer prediction using structured tabular data, primarily due to the absence of explicit temporal dependencies, which are unsuitable for sequence-based modeling. This work presents a novel approach that redefines how LSTM architecture can be applied to the publicly available non-sequential Wisconsin Diagnostic Breast Cancer (WDBC), consisting of 569 samples and 30 features. The flat tabular input is reshaped into a fixed-length 3D tensor using a sliding window approach to adapt the data for sequence modeling. This transformation enables the model to leverage LSTM's sequential processing capabilities in a fundamentally new way, capturing implicit feature interactions across structured attributes without temporal context. Furthermore, Bayesian hyperparameter optimization techniques are applied to enhance the model's performance. The proposed model is evaluated against standard LSTM and state-of-the-art tabular Transformer architectures (FT-Transformer and SAINT). Results show that BiLSTM achieves the best overall performance (AUC 0.9985, accuracy 0.9824, RMSE 0.0964), while the LSTM baseline also surpasses both Transformerbased tabular models (AUC 0.9958, accuracy 0.9719). Performance gains are consistent across seven evaluation metrics, with statistical significance confirmed via paired t-tests $({p}<0.05)$. These findings demonstrate that, when appropriately adapted, recurrent architectures can outperform even advanced self-attention models in structured clinical prediction tasks.

Immunotherapy with immune checkpoint inhibitors (ICI) has become a transformative pillar in cancer treatment, offering significant improvements in survival and reducing treatment-related side effects compared to traditional therapies. In gynecologic cancers, ICIs have transformed the treatment of endometrial (EC) and cervical cancers, whereas they have not demonstrated clinical benefit in ovarian cancer. This review examines the current state of ICI advancements in EC. Given the unique immunological characteristics of EC, a comprehensive understanding of advancements is crucial for optimizing decision-making and patient outcomes. While ICIs have demonstrated robust and durable efficacy in dMMR/MSI-H EC, the magnitude of benefit in pMMR disease remains modest. Additionally, we examine promising future directions, including personalized immunotherapy approaches and novel combination therapies (e.g. antibody-drug conjugates, PARP inhibitors, antiangiogenic drugs).

Orthogonal signal generators are crucial for synchronization in single-phase systems, where accurate estimation of phase, frequency and amplitude is the focal point. Conventional generators are sensitive to a DC-offset in the input signal, which can degrade performance. This paper presents a modified Kalman-based generator with an additional feedback loop for DC elimination. A state-space model of proposed generator is developed, and parameters are calculated using a continuous Kalman estimator. The performance is validated in MATLAB/Simulink environment under several tests to determine performance of the presented orthogonal signal generator. Simulation results show that the generator is accurately tracking the input signal while generating its quadrature components demonstrating robust performance suitable for synchronization loop applications.

Chainsaw felling and processing work is conducted in various natural conditions and requires significant physical effort from the workers, movement in severe weather and environmental conditions, and has a high risk of injury. The aim of this study was to determine the physiological workload of chainsaw operators through continuous heart rate measurement during the entire working day. The research was carried out during the summer of 2024, encompassing different parts of the Federation of Bosnia and Herzegovina. Heart rate was measured using a Polar H10 Heart Rate Monitor Chest Strap with continuous data logging and storage of heart rate readings. A time study was performed based on recordings conducted simultaneously with the recording of heart rate, with the aim of determining the duration of individual work operations and identifying the work operation with the highest negative impact on the worker. The average working heart rate during productive work time for subject 1 was 104 bpm, 83 bpm for subject 2, 109 bpm for subject 3, 94 bpm for subject 4 and 129 bpm for subject 5. The results of the Kruskal-Wallis test showed a statistically significant difference in average heart rate in relation to the time study element. The heart rate reserve (%HRR) for the whole study time was estimated at 41.05 % for subject 1; 22.69% for subject 2; 44.50% for subject 3; 24.04% for subject 4, and 45.78% for subject 5. The results of the study showed that the %HRR of chainsaw operators during felling and processing exceeded the value of 40% for 3 out of 5 subjects, which corresponds to hard work and may have negative consequences for operators´ health.

With the growth of data volume and increased query complexity, the need for the application of various optimisation techniques that enable faster execution and more efficient use of resources is increasingly becoming evident. Research shows that indexing, query execution optimisation, and the use of caching significantly reduce processing time and increase system responsiveness. Given that databases are constantly growing in size due to the need to store and analyse data, efficient database architecture and organisation are imperative to the business environment. This paper deals with the topic of analysing databases with large data sets and how to retrieve them most efficiently, using web applications, which are today the most common UI for databases.

C-reactive protein structurally belongs to the pentraxin family, calcium-binding proteins with immune defense properties. In the serum of healthy adults and adolescents, there is less than 5 mg of C-reactive protein. Its concentration is increased in inflammatory diseases where values up to 500 mg/l can be found. The main role of C-reactive protein is complement activation and prevention of inflammation. It binds to bacteria or damaged cells and thus helps the activation of the classic complement pathway, opsonization and phagocytosis. Binding depends on calcium. Antibiotics are products of the metabolism of bacteria, fungi and molds, rarely higher plants, which in small concentrations prevent the growth and development of microorganisms or kill them. They belong to the group of antimicrobial drugs, which are used to treat and prevent bacterial infections. Cephalosporins are beta-lactam antibiotics with the same mechanism of action as penicillin, which means that they block the synthesis of the bacterial cell

Atrial fibrillation (AF) is the most common persistent cardiac arrhythmia in clinical practice and a significant, often underdiagnosed risk factor for stroke. The electrocardiogram (ECG) is the primary method for its detection, typically manifesting as irregular $\mathbf{R R}$ intervals and the absence of P-waves. Numerical ECG parameters enable quantitative analysis of these changes and provide a foundation for the development of automated detection systems. This study examines the association between atrial fibrillation and numerical ECG parameters using the ECG-ViEW II database. From 12-lead ECG recordings, key temporal and morphological parameters were extracted, and descriptive statistics were calculated to form the final dataset. Descriptive statistical analysis, inferential tests, and graphical visualizations were applied to compare AF and non-AF groups. The results indicate that parameters describing RR-interval variability show a strong association with atrial fibrillation, confirming their potential for application in automated systems for early AF detection.

The recent rise of large language models (LLMs) and other generative artificial intelligence (AI) tools like agents represent only the visible tip of the iceberg in artificial intelligence. Beneath the surface lies a vast foundation of advanced mathematics, statistics, signal processing techniques, countless smaller/domain-specific models, optimization algorithms, distributed computing infrastructure, specialized engineering tools, and domain knowledge that make these headline-grabbing AI systems possible. This paper uses the iceberg metaphor to illuminate these hidden layers and surveys the technical background, including the mathematical and statistical underpinnings of AI, but also limitations of LLMs in many engineering applications such as power system fault detection, simulation-based optimization, and time-series forecasting. By exploring these “submerged” components of the AI iceberg, this paper provides a comprehensive perspective on the true breadth of modern AI, bridging the gap between public perceptions of AI and the complex reality beneath with multiple supporting case studies. The findings contribute to the existing body of knowledge by underscoring that meaningful progress in AI requires not only visible breakthroughs in models and interfaces, but also continuous advances in the less glamorous but critical supporting layers and applications of the AI stack.