Search Results (6,028)

Imbalanced datasets, where one class significantly outnumbers others, remain a persistent challenge in machine learning, often biasing predictions toward the majority class and degrading classifier performance. This paper provides a comprehensive, systematic review of data balancing methods, extending beyond foundational oversampling techniques such as the Synthetic Minority Oversampling Technique (SMOTE) and its variants (e.g., Borderline SMOTE, K-Means SMOTE, and Safe-Level SMOTE) to encompass advanced adaptive methods (MWMOTE, AMDO), deep generative models (generative adversarial networks, variational autoencoders, and diffusion models), undersampling techniques (NearMiss, Tomek Links), combination/hybrid methods (SMOTE-ENN, SMOTE-Tomek, and SMOTE+OCSVM), ensemble strategies (SMOTEBoost, RUSBoost, Balanced Random Forest, and One-Sided Selection), and specialized approaches for multi-label and clustered data. Beyond descriptive categorization, this review critically examines each method’s underlying assumptions, operational mechanisms, and suitability for diverse data characteristics, including high dimensionality, mixed feature types, class overlap, and noise. Key findings demonstrate that no single method universally outperforms others; optimal selection depends critically on dataset characteristics, classifier choice, and evaluation metrics. The paper concludes by identifying emerging research directions, including self-supervised learning for imbalance, diffusion-based generative oversampling, distribution-preserving resampling, knowledge distillation for imbalanced deployment, and the adaptation of foundation models to skewed distributions, offering practical guidelines for practitioners and a roadmap for future methodological development. Full article

Small non-coding RNAs (sRNAs) are ubiquitous post-transcriptional regulators that enable rapid bacterial adaptation to fluctuating environments. Previous reviews have largely focused on sRNA mechanisms in model organisms. This review integrates computational prediction, meta-omics-based discovery, and synthetic biology applications of small regulatory RNAs in marine and environmental prokaryotes, providing a multi-layered perspective from identification to functional and engineering applications. The current landscape of sRNA identification tools is critically evaluated, with emphasis on strategies to overcome challenges such as false-positive predictions. Recent advances in mapping the RNA interactome and emerging evidence of previously underappreciated roles of sRNAs in environmental adaptation are discussed. Additionally, metagenomic and metatranscriptomic studies revealing the diversity of environmental sRNAs in uncultured microbial communities are summarized, highlighting their ecological significance. Finally, a curated overview of synthetic sRNA applications in metabolic engineering, including target genes and enhanced product yields, is provided as a resource for strain engineering. Collectively, this review provides a holistic view of prokaryotic sRNA biology, distinguishing it from more narrowly focused studies. Overall, sRNAs are highlighted as key regulatory elements linking microbial environmental adaptation with emerging biotechnological applications through advances in meta-omics guided discovery and synthetic RNA engineering. Full article

Background: Intelligent decision support systems for colonoscopy can fail when encountering out-of-distribution surgical instruments such as snares or biopsy forceps, producing false-positive polyp masks that may contribute to automation bias and reduce workflow reliability. This study aimed to develop and evaluate a parameter-efficient framework for open-set-aware polyp segmentation that can reject such anomalous inputs while preserving in-distribution segmentation performance. Methods: We propose PolypSAM-Open, which integrates a prototype-based Open-Set Learning (OSL) module with Low-Rank Adaptation (LoRA) in the MedSAM image encoder. The model was trained on Kvasir-SEG using an 85/15 split of authentic polyp images and synthetic high-frequency Gaussian noise to learn a rejection margin. Zero-shot out-of-distribution detection was evaluated on 590 unseen authentic surgical instruments from Kvasir-Instrument. Segmentation and detection performance were compared against a standard MedSAM-LoRA baseline. Results: Standard parameter-efficient fine-tuning yielded an OOD AUROC of 0.4263 on authentic surgical instruments. PolypSAM-Open improved zero-shot OOD AUROC to 0.9535 ($p<0.001$). Despite allocating 15% of training capacity to the synthetic-noise rejection margin, PolypSAM-Open maintained segmentation performance comparable to the standard fine-tuned baseline (Dice 0.9728 versus 0.9723). On ETIS-LaribPolypDB and CVC-ClinicDB, Dice scores were 0.9301 and 0.9386, respectively. The approach remained parameter-efficient, updating 4.48% of total parameters while adding negligible inference latency relative to the underlying MedSAM forward. Conclusions: Prototype-based open-set adaptation can substantially improve rejection of unseen surgical artifacts in AI-assisted colonoscopy while preserving high segmentation accuracy. These findings position PolypSAM-Open as a promising strategy for potentially safer decision-support segmentation in endoscopic workflows; prospective clinical validation remains necessary. Full article

Smart sensor networks for environmental monitoring require accurate and continuous estimation of key variables such as temperature, humidity, and wind speed; however, physical sensor deployments are frequently limited by high costs, hardware failures, and data quality degradation, while existing virtual sensor approaches rely on single-model architectures that lack explicit uncertainty modeling and fail to capture the complex non-linear dynamics of real-world IoT time-series data. This paper proposes VAS-DPFF, a virtual augmented sensor framework based on deterministic and probabilistic feature fusion, which contributes a principled integration of well-established deterministic and probabilistic techniques within a unified AIoT-compatible virtual sensing architecture. The framework integrates: (i) a deterministic pipeline comprising temporal encoding, rolling statistics, and mutual information-based feature selection; (ii) a probabilistic pipeline employing Bayesian Ridge Regression (BRR) and Gaussian Process Regression (GPR) to generate uncertainty-aware synthetic features; and (iii) an early feature-level fusion strategy feeding an XGBoost regression model augmented with Gaussian noise injection. Experiments on 84,582 time-series records from a nine-station IoT environmental monitoring network in Gwacheon City, South Korea, demonstrate strong multi-target prediction performance: temperature RMSE $=0.811$ ${}^{\circ }$C, R${}^2=0.973$; humidity RMSE $=4.113$%, R${}^2=0.964$; and wind speed RMSE $=0.602$ m/s, R${}^2=0.798$, representing RMSE reductions of 61.2%, 60.7%, and 62.3% over the existing method, respectively. Comprehensive ablation studies, sensitivity analysis, and augmentation validation confirm that the proposed integration of deterministic and probabilistic features yields consistent and practically valuable improvements in virtual sensing performance suitable for AIoT-enabled smart sensor network deployments across multiple environmental monitoring targets. Full article

The degradation of organic pollutants through photocatalysis is currently a major research focus. Core–shell heterostructures of metal semiconductors have been widely recognized as an effective strategy for enhancing photocatalytic performance, particularly when alloy nanoparticles are incorporated due to their unique electronic and catalytic properties. However, conventional synthetic approaches typically rely on high-temperature and high-pressure conditions, which often induce undesirable particle overgrowth and aggregation. Herein, AuPt bimetallic alloy nanoparticles were successfully fabricated via two successive in situ redox processes under room-temperature and ambient-pressure conditions, which were in situ integrated with Cu₂O to form multilevel core–shell composite particles. Structurally, an octahedral Cu₂O crystal serves as the inner core (denoted as Cu₂O_(O)), sequentially coated with a Cu₂O nanoparticle (denoted as Cu₂O_(P)) interlayer and a AuPt alloy nanoparticle shell. Functionally, the enhanced photocatalytic activity of Cu₂O_(O)@Cu₂O_(P)@AuPt was proven to be attributed to a dual function of AuPt, which includes an adsorption-induced polarized interface and an efficient charge-transfer mediator with the ohmic contact. This work demonstrates a mild and versatile synthetic strategy for constructing semiconductor–alloy heterostructures, offering valuable insights into the rational design of highly efficient and stable photocatalysts. Full article

An innovative strategy for collagen self-assembly with accelerated intra and extrafibrillar mineralisation is introduced to generate bone scaffolds with biomimetic properties. This method, termed Rapid Fibrillogenic Mineralisation (RFM), leverages coprecipitation with 10× Simulated Body Fluid (10× SBF) during fibril formation to maximise nucleation, particularly within intrafibrillar zones at molecular termini. Densification is achieved within minutes via plastic compression driven by capillary action, producing bone-like scaffold density without compromising the collagen matrix. Transmission electron microscopy confirms intrafibrillar hydroxyapatite crystals within 15 min, while X-ray diffraction demonstrates distinct HA peaks across groups. Scanning electron microscopy verified extrafibrillar mineralisation after 4 h, with saturation by 6 h, yielding ‘nanoflower’ crystal clusters. Infrared spectra showed increased carbonate content over time, indicating lattice substitutions characteristic of natural bone. Enhanced mineralisation translated into significant mechanical gains as Dynamic Mechanical Analysis revealed compressive moduli approaching cancellous bone (up to 283 ± 31 MPa). In addition, a decrease in the piezoelectric coefficient occurs with increased mineralisation process, highlighting the effects of mineral inclusions on collagen fibre composition and anisotropy. Biologically, mineralised scaffolds supported cellular growth compared to collagen controls. RFM thus enables rapid, reproducible fabrication of biomimetic bone scaffolds that closely emulate native mineralisation patterns and mechanical behaviour. Beyond offering a practical route for scaffold production in tissue engineering, the process also provides new insights into bone physiology and in vitro modelling. By reshaping collagen into a synthetic echo of nature’s bone, RFM establishes a rapid approach for designing functional biomaterials with translational potential. Full article

This paper proposes a federated learning framework for privacy-preserving patient monitoring with lightweight blockchain anchoring. The framework keeps synthetic patient monitoring records local at each client and uses federated model aggregation to support collaborative learning without centralizing raw records. To improve traceability, the blockchain layer is specified as an anchoring mechanism that records compact evidence, including model hashes and participation metadata, rather than raw data or full model parameters. Experiments were conducted on synthetic patient monitoring data to evaluate framework behavior under non-IID client distributions, label noise, different client counts, partial client participation, and aggregation strategies. The centralized MLP baseline achieved approximately 0.89 overall accuracy but failed to detect alert cases, with 0% alert-class recall, showing that accuracy alone can be misleading in imbalanced monitoring scenarios. In the federated simulations, the model reached approximately 0.99 accuracy under clean labels, approximately 0.90 under 10% label noise, and approximately 0.70 under 30% label noise. Under a more difficult noisy, non-IID, dropout, and fixed skewed-client evaluation setting, the model stabilized at approximately 0.80 accuracy after 25 communication rounds. Client scaling from 5 to 20 clients remained stable, and FedAvg, weighted aggregation, and accuracy-trimmed robust aggregation produced similar final accuracy of approximately 0.98 in the 10-client setting. The results indicate that label quality strongly affects federated convergence, while blockchain anchoring should be interpreted as an auditability mechanism rather than a direct accuracy-improving component. This study provides a framework-level foundation for auditable federated patient monitoring in semi-trusted healthcare networks. Full article

Drug discovery is a complex and resource-intensive process, with lead identification representing a major bottleneck due to high attrition rates. Natural products have long served as a valuable source of structurally diverse and biologically active compounds, offering advantages such as evolutionary optimization, target specificity, and unique chemical diversity. This review provides a comprehensive and mechanistic overview of natural products as lead compounds, emphasizing their chemical characteristics, biological relevance, sources, mechanisms of action, and integration into modern drug discovery pipelines. A narrative review was conducted using major scientific databases (PubMed, Scopus, Web of Science, and Google Scholar), covering literature from 2000 to 2026. Relevant studies were selected based on scientific rigor and contribution to key themes, including natural product diversity, discovery strategies, and technological advancements. Natural products exhibit superior structural complexity and occupy unique chemical space compared to synthetic compounds, enabling effective interaction with diverse biological targets and supporting polypharmacological activity. Key sources include plants, microorganisms, and marine organisms, which have yielded numerous clinically important drugs. Advances in analytical techniques, genome mining, metabolomics, synthetic biology, and artificial intelligence have significantly improved discovery and optimization processes. Despite challenges related to complexity and scalability, natural products remain indispensable in drug discovery, with emerging technologies enhancing their potential for addressing unmet medical needs. Full article

The cysteine-based thiol–epoxy ‘click’ reaction is demonstrated as an efficient and practical approach for the synthesis of zwitterionic structures. The transformation employs unprotected cysteine, proceeds in aqueous media, and affords quantitative conversions. Notably, acid- and base-catalyzed conditions provide exclusive access to different cysteine-based thioether regioisomers in aqueous conditions. The pH-responsive behavior of the resulting zwitterions is further elucidated by NMR spectroscopy. Finally, the synthetic strategy is extended to the preparation of cysteine- and glutathione-functionalized polyethylene glycol polymers, showcasing its utility for the preparation of amino acid-/peptide-containing macromolecular materials. Full article

Extracellular vesicles (EVs), particularly small EVs or exosomes, are promising cell-free therapeutics with superior biocompatibility and intrinsic targeting for synthetic nanoparticles. However, conventional bulk preparation methods suffer from low yield, poor reproducibility, and structural instability. Microfluidic technologies resolve these issues by enabling precise, automated, and low-shear fluidic manipulation. This mini-review highlights recent advances in microfluidic-engineered exosomes for cancer immunotherapy and infectious diseases. We evaluate critical microfluidic strategies for isolation, surface engineering, and cargo loading, contrasting platforms like ExoArc, acoustofluidics, cellular nanoporation, and electroporation. Particular emphasis is placed on complex modalities, including immune cell-derived exosomes (IEX), neo-antigen presentation, chimeric antigen receptor (CAR)-derived exosomes, and targeted siRNA delivery networks. Crucially, we analyze the technological disconnect between analytical microfluidic scales and massive therapeutic manufacturing volumes, addressing how physical forces risk damaging conformationally sensitive surface proteins (e.g., CAR scFv). Finally, we outline future perspectives, including high-throughput 3D-multiplexed networks, stimulus-responsive scarless elution, and integrated “sample-to-therapy” circuits. Guided by the MISEV2023 guidelines, this review frames the path toward standardized, clinical-scale engineering of multi-functional, cell-free immunotherapies. Full article

Color is a critical sensory attribute of foods that strongly influences consumer perception, acceptance, and purchasing decisions. As health-oriented consumption increases, natural pigments are progressively replacing synthetic colorants. Red algal phycobiliproteins (PBPs) have gained attention in related industries for their vivid water-soluble colors and potential health-promoting properties. Phycoerythrin (PE), phycocyanin (PC), and allophycocyanin (APC) are key pigment proteins responsible for their characteristic coloration. However, variations in algal species, season, and cultivation environments make PBP composition, yield, and quality difficult to standardize. This review summarizes red algal PBPs, covering algal sources, extraction and purification, physicochemical properties, biological functions, regulatory frameworks, and future directions. Major red seaweed sources, PBP types, and extraction strategies are compared, with cyanobacterial PBP studies incorporated where direct red algal evidence is limited. Evidence suggests that red algal PBPs are promising for clean-label, water-based, and mildly processed foods, including beverages, dairy products, confectionery, and meat alternatives, but their application is constrained by sensitivity to heat, light, oxygen, and pH. Beyond coloration, PBPs exhibit antioxidant, anti-inflammatory, anticancer, antibacterial, and metabolic health-promoting activities. Overall, red algal PBPs have considerable potential as dual-function ingredients, although commercialization requires advances in raw material standardization, stability-enhancement strategies, process optimization, clinical validation, and regulatory harmonization. Full article

Tool micro-defect quantification is often degraded by optical sampling limits, boundary aliasing, and non-Gaussian industrial noise. To improve segmentation fidelity and physical measurement reliability, this study proposes a physically guided reconstruction and metrology framework for milling tool defects. The framework first uses an improved ResNet18-BiFPN segmentation network supervised by a distance transform-based boundary-aware loss, which encourages mask boundaries to fit high-curvature crack and chipping regions. A tensor-guided elliptical adaptive structuring element is then introduced to restore local topology while preserving tangential connectivity and normal boundary fidelity. Finally, a stable edge region search and curvature-adaptive gray/Zernike moment fusion strategy are used for noise-robust sub-pixel localization and physical area quantification. Image-based evaluation and controlled simulation-based stress tests show that the proposed method achieves an mIoU of 92.5%, an F1-score of 97.1%, and an HD95 of 2.15 pixels. Under strong synthetic noise, the average distance error remains below 0.28 pixels. For micro-defect area measurement, the mean relative error is reduced to approximately 1.7%. These results indicate that the proposed framework can support more reliable defect metrology for tool condition monitoring under complex industrial imaging conditions. Full article

Soil salinity is a major constraint to agricultural productivity, causing osmotic stress, ion toxicity, nutrient imbalance, and progressive deterioration of soil biological functions. Beyond its direct effects on plant performance, salinity also generates persistent soil legacies that influence subsequent plant growth through plant-soil feedback (PSF) processes. PSF provides an ecological framework for understanding how plants modify the physicochemical and biological properties of soil and how these altered soil conditions subsequently affect plant growth, health, and resilience. Salinity research has predominantly emphasized soil microorganisms as promoters of plant growth, while their broader role in regulating soil legacy effects remains comparatively underexplored. This review examines whether soil microorganisms may contribute to a transition from salt-amplified negative PSF toward more favorable feedback outcomes by reshaping rhizosphere chemistry, nutrient cycling, pathogen pressure, ion homeostasis, stress signaling, and soil structural stability. However, conditioned-soil bioassays and multi-season saline field trials remain scarce, these proposed pathways are treated as potential mechanisms or testable hypotheses rather than as established evidence of PSF regulation. We first summarize the mechanisms underlying PSF in non-saline systems and then describe how salinity alters plant-, soil-, and microbe-mediated feedback pathways. We further evaluate the potential of halotolerant plant growth-promoting rhizobacteria, arbuscular mycorrhizal fungi, actinobacteria, disease-suppressive microbial communities, and synthetic microbial consortia as regulators of PSF, while distinguishing direct salt-tolerance effects from evidence of genuine feedback modulation. Specifically, improved salt tolerance in the inoculated plant is interpreted as direct stress mitigation, whereas demonstrated PSF regulation additionally requires measurable soil conditioning and an effect on a subsequent crop. The novelty of this review lies in organizing studies of salinity-microbiome interactions within an evidence-based PSF framework that differentiates immediate plant responses from rhizosphere modification, conditioned-soil effects, and subsequent-crop performance. The review concludes that microbial strategies for saline agriculture are most likely to succeed when developed as integrated PSF interventions that combine crop traits, indigenous microbiomes, optimized inoculant design, organic matter management, diversified rotations, and multi-season field validation. Full article

Global food demand is projected to rise by approximately 56 percent between 2010 and 2050, intensifying reliance on synthetic nitrogen fertilizer during a time when only about half of all applied nitrogen is recovered by crops, with the remainder split between genuine environmental loss and retention within soil and biomass pools. The fraction that is genuinely lost drives substantial economic costs and contributes disproportionately to global nitrous oxide emissions, a greenhouse gas with a warming potential far exceeding that of carbon dioxide. This review synthesizes recent literature across three interdependent domains including nitrogen use efficiency strategies spanning agronomic, genetic, and microbial approaches, decarbonization pathways for ammonia synthesis ranging from conventional to green production routes, and gas sensing technologies for monitoring ammonia and nitrous oxide emissions in agricultural settings. Rather than treating these domains separately, this review proposes that their effects on overall emissions are complementary and potentially compounding rather than strictly additive. Efficiency improvements reduce the total fertilizer volume subject to production emissions, while cleaner production cannot offset nitrogen loss in the field. The exact extent of any combined benefit depends on the relative proportion of field emissions and production emissions within each farming system. Another important finding is the pronounced asymmetry in monitoring readiness. Ammonia sensing has reached field-deployable maturity for detection and concentration monitoring. In contrast, nitrous oxide sensing remains constrained by unresolved challenges in sensitivity and long-term stability despite the gas’s significant contribution to climate change. This asymmetry limits the verification of mitigation outcomes at farm and regional scales. The review further identifies that intervention effectiveness depends on farm structure in the studied context, that global nitrogen policy remains weighted toward incentivizing use rather than reducing pollution, and that the evidence base surveyed here is geographically uneven. Together, these findings indicate that reconciling rising food production with greenhouse gas reduction targets requires integrated frameworks linking field nitrogen budgets, production emissions, and monitoring capability, alongside policy instruments designed around their interdependence. Full article

This review summarizes the synthesis and structural modification of α-pyrone-containing phloroglucinol derivatives from Helichrysum species. Synthetic routes to both natural products and synthetic analogues are covered, highlighting strategies ranging from β-keto ester cyclodehydration to multicomponent condensations. Key methods include aldehyde-mediated dimerization, fluoride-catalyzed heterodimerization, and acid-catalyzed cyclization to benzopyran frameworks. The reported approaches enabled access to diverse monopyrone, dipyrone, arzanol-type, and cyclized analogues. Additionally, retrosynthetic analyses toward phloroglucinol–pyrone heterodimers are discussed, highlighting convergent synthetic strategies for future access to benzofurane-, chromene-, and chromane-based analogues. This review integrates published experimental data with newly generated in silico predictions. Full article