Search Results (10,333)

The growing problem of air pollution and its direct impact on human health have created an urgent need for reliable, intelligent, and machine learning (ML)-enabled gas-sensing technologies. Among various sensing platforms, metal oxide gas sensors (MO-GSs) have emerged as promising candidates owing to their low cost, high sensitivity, and scalability. However, their practical application is limited by poor selectivity, cross-sensitivity, sensor drift, and high operating temperatures. Recent advances in ML have provided effective strategies to overcome these limitations through data-driven optimization of sensing performance. This review summarizes recent progress in ML-assisted MO-GSs, covering sensor array design, feature engineering, and classification algorithms, including support vector machines (SVMs), random forests (RFs), and deep neural networks (DNNs). In addition, key data-processing techniques such as preprocessing, dimensionality reduction, and hybrid learning approaches are critically discussed. The application of ML-enabled MO-GSs in medical diagnostics, environmental monitoring, industrial safety, and food quality assessment is also reviewed. Despite significant progress, challenges including limited dataset availability, sensor drift, and poor model generalization remain. Future research should focus on developing adaptive, energy-efficient, and IoT-enabled smart sensing systems. The integration of machine learning with metal oxide gas sensors represents a significant step toward intelligent, next-generation, high-performance gas-sensing technologies. Full article

With the increasing penetration of wind and photovoltaic generation, microgrids are playing an increasingly important role in promoting renewable energy accommodation, enhancing operational flexibility, and enabling low-carbon energy management. However, the strong uncertainty of renewable generation and load demand, together with the coupling effects of electricity, hydrogen, and carbon markets, poses significant challenges to the optimal planning and operation of microgrid energy storage systems. To address these issues, this paper proposes a bi-level optimal sizing framework for an electric–hydrogen hybrid energy storage system (EHH-ESS) in a microgrid under multi-market coupling. First, typical wind–solar–load scenarios are generated using a Wasserstein generative adversarial network with gradient penalty (WGAN-GP), so as to capture the stochastic characteristics and temporal correlations of renewable generation and load demand. Then, a multi-market coupling index (MCI), integrating electricity price, hydrogen price, and carbon price signals, is constructed to characterize time-varying economic and low-carbon operating incentives and to guide coordinated dispatch decisions. On this basis, a bi-level multi-objective optimization model is established. The upper level determines the optimal capacities of battery storage, electrolyzers, fuel cells, and hydrogen tanks, while the lower level performs hourly coordinated operation of the microgrid under multi-market conditions. The model considers annual equivalent total cost, renewable energy curtailment rate, and carbon emissions as objective functions, and is solved using the NSGA-III algorithm. Compared with the no-storage benchmark, the proposed scheme improves the annual operating economics and renewable-energy accommodation under the studied market conditions. The proposed method significantly reduces annual operating cost and improves renewable energy accommodation. However, under the current carbon price and grid emission factor settings, the optimal economic solution increases carbon emissions relative to the baseline, indicating a trade-off between economic arbitrage and low-carbon operation. Full article

Drug discovery is still an expensive and time-consuming process where finding the right drug associations is important for therapeutic development. In this paper, a new system is proposed for drug design called PharmaGraphFormer (PGF). It consists of five stages: (i) Data acquisition and preprocessing (DAP), (ii) Feature extraction and feature fusion (FEF), (iii) Molecular representation (MR), (iv) Multi-task prediction, and (v) Explainable artificial intelligence (XAI). This study employs a hybrid graph neural network (GNN)-transformer architecture that combines structural and sequence-based representations. Through DAP, several processes are executed, including the imputation or removal of missing values, outlier rejection, and class balancing. Next, through FEF1, features are extracted to represent the input data efficiently. Initially, compound-protein features are generated to document the interactions and relationships between chemical compounds and their corresponding target proteins. Secondly, drug characterizations are computed to encapsulate the physical, chemical, and structural attributes of each drug. After that, MR is performed using a graph-based molecule representation. Then, a novel model integrating GNNs and graph transformers, termed GNN-T, is proposed. Initially, GNNs represent the most promising deep learning models adept at processing non-Euclidean data. The Graph Transformer layer enhances atom representations by consolidating the representations of adjacent atoms through an attention mechanism. Finally, XAI is applied to explain the internal mechanisms of AI systems, rendering them comprehensible and interpretable. Across five independent runs, the proposed model achieved an accuracy of $0.963\pm 0.002$, a precision of $0.971\pm 0.002$, a recall of $0.958\pm 0.003$, an F1-score of $0.964\pm 0.002$, and a ROC-AUC of $0.993\pm 0.001$. These results demonstrate an outstanding performance when compared with all other models and emphasize that the proposed model is reliable in solving the problems of prioritizing compounds in line with the latest developments in AI-powered virtual screening and drug–target interaction modeling. Full article

Background/Objectives: Chimeric antigen receptor (CAR)-T cell therapy has achieved substantial clinical success in haematological malignancies but has shown limited efficacy against solid tumours. Key barriers include inadequate tumour trafficking, immunosuppressive tumour microenvironments, poor selectivity and heterogeneity of antigen expression, and challenges related to safety and manufacturing. Biomaterial-based technologies have emerged as a potential strategy to address many of these limitations. This review aims to critically evaluate biomaterial approaches designed to enhance CAR-T cell therapy of solid tumours and assess their translational potential. Methods: A narrative review of recent pre-clinical translational studies was conducted, focussing on biomaterial platforms developed to improve CAR-T cell delivery, persistence, functionality, safety control, and manufacturing efficiency in solid-tumour settings. Approaches were analysed according to their mechanisms of action, therapeutic benefits, and stage of translational readiness. Results: Biomaterial strategies, including nanoparticles, injectable and implantable hydrogels, scaffolds, and hybrid delivery systems, have improved CAR-T infiltration, survival, and therapeutic efficacy in several solid-tumour models. Localised delivery of cytokines and other immunomodulatory cues enabled improved spatio-temporal control of CAR-T activation, reducing systemic toxicity, and increasing persistence. Additional applications include amplified ex vivo CAR-T expansion and support for non-viral or in vivo CAR-T generation. However, increased material complexity was frequently associated with challenges in scalability, regulatory approval, and long-term safety. Conclusions: Biomaterial-enabled approaches offer a versatile toolkit to address key biological and translational barriers limiting CAR-T cell therapy of solid tumours. Strategies based on clinically familiar materials and simplified designs appear most suitable for near-term clinical translation, emphasising the need to balance engineering innovation with safety, scalability, and integration into existing clinical workflows. Full article

Mobile robots and autonomous vehicles require tightly regulated direct current (DC) power under rapidly varying load conditions, motivating control strategies that combine fast nonlinear regulation with predictive optimization. This paper proposes a Hybrid Sliding Mode Control and Model Predictive Control (Hybrid SMC + MPC) strategy for a DC-DC buck converter supplying a representative mobile-robot mission load. The controller employs a cascade SMC structure for fast inner-loop regulation and an MPC component that provides finite-horizon duty-cycle correction using planned load information. The MPC problem is formulated in condensed form and solved analytically without an external optimization solver. A Lyapunov-based analysis establishes a sufficient reaching condition for the sliding variable under the ideal averaged-model assumptions, and the condition is verified for the simulated mission. The proposed approach is evaluated in MATLAB using a 10-phase, 10 s load profile with resistance varying from 7 Ω to 100 Ω and is compared with SMC-only, MPC-only, PID, constant-duty, and reconstructed fuzzy-logic benchmarks. In the averaged-model study, the Hybrid SMC + MPC achieves a maximum absolute voltage deviation of 0.388 V, an RMSE of 0.0115 V, and a final-phase mean absolute error of 0.0076 V. It provides the lowest maximum voltage deviation among the principal closed-loop controllers, while PID achieves the lowest RMSE and final-phase error and SMC-only exhibits the shortest mean settling time. Relative to MPC-only, the Hybrid controller reduces the maximum voltage deviation by approximately 43.6% and the mean settling time by approximately 66.1%. An ablation study shows that the MPC contribution substantially improves overall and steady-state regulation accuracy, while load preview primarily reduces the worst-case voltage deviation. Switching-level MATLAB/Simulink validation with explicit 20 kHz PWM and converter parasitics confirms that the output remains within ±2% of the 25 V reference throughout the complete mission, with a maximum absolute deviation of 0.443 V and a maximum steady-state switching ripple of 21.6 mV peak-to-peak. These results demonstrate that the proposed Hybrid SMC + MPC architecture provides a favorable balance between worst-case transient regulation, steady-state accuracy, and predictive control capability for dynamically varying robotic power loads. Full article

Urban residents spend an estimated 80–90% of their time indoors, yet urban indoor air quality (IAQ) science remains fragmented across pollutant types, settings, mitigation strategies, and governance contexts, contributing to an estimated 6.7 million deaths annually from indoor air pollution worldwide. This review asks the following question: what are the critical, multi-dimensional research and governance gaps in urban IAQ, and how can they be systematically derived and organized into a reference framework for future research and policy? A PRISMA 2020-aligned hybrid systematic evidence synthesis, combining bibliometric science mapping and structured thematic synthesis, was conducted across 105 records, primarily published between 2011 and 2026, with three pre-2011 foundational records retained, spanning 15 national contexts. A seven-stage hybrid deductive–inductive derivation procedure was applied to the coded corpus to produce the Multi-Dimensional Gap Identification Framework (MGIF), organizing research gaps across five dimensions: knowledge, methodological, technological, policy and implementation, and equity and urban context. Recurrent gaps include the absence of multi-pollutant mixture assessment in monitoring frameworks, the lack of standardized measurement protocols limiting cross-study comparability, a systematic gap between laboratory-validated and field-measured intervention performance, the near-total absence of enforceable indoor air quality standards across most jurisdictions, and the severe underrepresentation of Global South populations in both primary evidence and regulatory design. Building on the MGIF output, the Urban Indoor Air Quality Nexus (UIAQN) is proposed as a four-level conceptual organizing architecture linking pollutant dynamics, building systems, personal exposure, and governance mechanisms. Both frameworks are grounded in the coded corpus, have not been subjected to external validation, and are designed as structured reference architectures for future research investment, standard harmonization, and equity-centered policy design rather than as empirically validated predictive models. Full article

To better understand how journalists represent sources, it is necessary to systematically study the use of reported speech in news coverage. This paper presents a method for LLM-assisted content analysis to identify and classify reported speech in Dutch newspapers automatically. The study evaluates a three-step procedure utilising role-based instructions to prompt the model as a professional journalist. First, a codebook for identifying citation structures and source types was developed with LLM support and then manually verified. Second, inter-coder reliability between human coders and the LLM was assessed on a representative sample of Dutch news articles using a human-in-the-loop validation approach. Third, the prompt-engineered LLM was used to code a large corpus spanning seven decades (1950–2024). Manual verification of 16,689 citations shows a weighted F1-score of 0.75, which aligns with recent benchmarks for high-capacity models performing complex journalistic coding. While human oversight remains the benchmark for reliability, due to issues such as repeated citations that were given as examples in the used prompts and representational bias, LLM-based systems perform sufficiently well for large-scale analyses of journalistic source use. The paper concludes that hybrid human–AI workflows provide a practical bridge between traditional rule-based approaches and new generative models, offering scalable and cost-effective methods for studying source representation in journalism. Full article

The Fuzzy Cognitive Map (FCM) framework provides a useful tool for representing the complex interdependencies involved in wastewater treatment selection, particularly when social, ecological, climatic, and economic criteria are considered simultaneously. In this study, the FCM approach was applied to compare two wastewater treatment approaches in Ahvaz, Iran: constructed wetlands (CWs) as a nature-based solution and energy-based wastewater treatment plants. The developed model included 30 components and 127 causal links, and was used to examine four scenarios representing CWs, energy-based treatment, a hybrid approach, and direct wastewater discharge. The results showed that both CWs and energy-based solutions had similar effects on public health, while the hybrid scenario produced the greatest improvement. CWs had a positive effect on ecosystem restoration and showed better performance in heavy metal removal, whereas energy-based solutions had a greater negative influence on environmental conditions and climate-related components. In addition, the economic results indicated that CWs were more favorable in terms of capital cost, energy consumption, and operational cost. Sensitivity analysis using ±10% variations in causal weights showed that the main scenario-response patterns remained generally unchanged. Overall, the findings suggest that CWs and energy-based systems each have specific advantages and limitations, while the hybrid approach offers the most balanced performance across the evaluated criteria. This study demonstrates the usefulness of the FCM approach for supporting wastewater management decisions and for identifying trade-offs among treatment alternatives in sustainable water resource planning. Full article

In the era of global energy transition, the development of third-generation photovoltaic technologies, such as dye-sensitized solar cells, has emerged as a paramount challenge in materials engineering. This study is dedicated to the synthesis and implementation of graphene quantum dots as eco-friendly sensitizers within DSSC architectures. The GQDs were synthesized via a microwave-assisted hydrothermal route using biodegradable organic precursors, providing a “green” alternative to conventional, toxic heavy-metal-based materials. The nanocrystalline structure and optoelectronic properties of the sensitizer were verified through UV-Vis and visual photoluminescence assessment. A focal point of this research was the optimization of the GQD concentration on the mesoporous surface of the titanium dioxide photoanode. Measurements were conducted utilizing a custom-designed experimental setup integrated with 3D-printed (FDM) components and an Arduino microcontroller, ensuring precise data acquisition under controlled illumination conditions (405–625 nm). The results indicated an optimal operational point at a fivefold dilution of the stock solution (0.4 g/dm³), which yielded the highest open-circuit voltage ($V_{\mathrm{oc}}$) of 545.4 mV under UV irradiation. The decline in photovoltaic performance observed at higher concentrations was attributed to excessive nanostructure agglomeration, which effectively blocked the mesopores of the semiconductor. Furthermore, the demonstrated high chemical capacitance of the system imparts electrochemical capacitor-like characteristics to the cell, enabling energy stabilization under fluctuating illumination. To elucidate the underlying sensitization mechanisms at the atomic level, computational simulations were conducted utilizing the MACE machine-learning potential and the GFN2-xTB semi-empirical method. The theoretical models revealed that the formation of stable covalent Ti–O–C bridges (chemisorption) is imperative for establishing strong interfacial electronic coupling. Solvation models and molecular dynamics (MD) at 300 K confirmed the thermodynamic and operational robustness of the hybrid system in an aqueous electrolyte. Ultimately, this combined experimental and theoretical work conclusively demonstrates that graphene quantum dots represent an efficient, highly stable, and non-toxic alternative to classic molecular dye sensitizers. Full article

Traditional rule-based intrusion detection systems generally fail in identifying unknown or evolving threats; thus, automated and adaptive kinds of methods are crucial. Deep learning models provide promising solutions, but many recent studies depend on hybrid architecture, which increase the computational cost and reduce deploying ability on real-time or resource-limited systems. In this paper, we present and test a standalone LSTM model for multiclass cyberattack detection based on a CIC_IoT_Dataset2023, a recent labeled dataset that mirrors the actual network environment containing 33 attack categories. The dataset was extremely imbalanced as benign traffic accounted for most of the classes. To detect such attacks, we used the Synthetic Minority Oversampling Technique (SMOTE) to increase the frequency of less common types of address. The pre-processed dataset was then employed to train four models (RNN, CNN, DNN and the proposed LSTM) for performance analysis with sequential data. The proposed LSTM model achieved an accuracy between 2% and 7%. LSTM had good detection for frequent attacks and slow-changing patterns, which shows its capacity in learning long-lasting dependencies. The results demonstrate that a simple, lightweight standalone LSTM model can be used for effective and realistic intrusion detection without the need for complex hybrid architecture. Full article

Context: Forage bermudagrass (Cynodon dactylon) is widely used in warm-season livestock production systems because of its high productivity and adaptability. However, systematic evaluation of forage-type germplasm remains limited, restricting the identification of superior breeding materials. Aims: This study aimed to evaluate phenotypic variation, identify key yield-related traits, and identify high-performing forage bermudagrass germplasm with contrasting interannual phenotypic responses derived from a ‘Wrangler’ × ‘CD-21’ hybrid population. Methods: Two evaluation populations were established. A single-genotype population of 621 individuals was used to assess plant and canopy height variation, whereas 16 representative entries were evaluated for biomass yield and major agronomic traits during 2024–2025. Frequency distribution, principal component, correlation, and path analyses were conducted. Key results: Stem height and canopy height showed unimodal, approximately normal distributions, indicating continuous phenotypic variation and supporting their characterization as quantitative traits. Biomass yield was positively associated with stem height (r = 0.79), canopy height (r = 0.82), and internode length (r = 0.63). Path analysis indicated that stem height had the largest estimated direct effect (β = 0.45) on biomass yield within the proposed path model. Multivariate analyses revealed distinct phenotypic differences among entries and years, allowing classification into high-performing, environmentally responsive, and leaf-structure efficient groups. Conclusions: Stem height, canopy height, and internode length were identified as key traits associated with forage biomass production. Integrating multivariate and path analyses effectively differentiated forage bermudagrass germplasm based on yield performance and agronomic traits. Implications: The identified germplasm and trait relationships provide useful information for further breeding evaluation and selection decisions and support the development of improved forage bermudagrass cultivars. Full article

Reliable retrieval-augmented generation requires consistency across query interpretation, evidence selection, and final answer generation. This study defines computational symmetry as bidirectional coverage among canonical query constraints, traceable evidence, and answer claims, with residual asymmetry triggering correction or abstention. The proposed framework integrates a source-linked raw text/entity/event knowledge graph, hybrid dense–sparse retrieval, cross-encoder reranking, pre-retrieval semantic alignment, and a post-retrieval verification gate. DeepSeek-V3 serves as the implementation backbone, while “ChatGPT-type” denotes the broader class of instruction-following conversational large language models. Experiments use T2Ranking for retrieval and reranking, ATIS for diagnostic intent–slot evaluation, and controlled dialogue scenarios derived from T2Ranking. The hierarchical representation improves retrieval F1 from 0.586 to 0.660, while the complete pipeline increases average answer correctness from 0.530 to 0.611 compared with direct LLM answering and from 0.559 to 0.611 compared with graph retrieval. On ATIS, the controller achieves 92.61% intent accuracy, below Joint BERT at 95.18%, and is therefore treated as a reusable orchestration module rather than a superior classifier. The results support the proposed verification correction framework within the tested settings, without claiming superiority over untested adaptive RAG systems. Full article

Geochemical gold datasets from shear zone-hosted systems frequently contain extreme outliers that distort statistical structure, shift population boundaries, and undermine the reliability of concentration–area (C–A) fractal modeling. Conventional treatments such as discarding anomalous samples or applying fixed Winsorization thresholds often fail to preserve the multivariate relationships that control geochemical dispersion. In this research, an intelligent random forest (RF)-based model was developed to reconstruct an extreme Au outlier in stream sediment samples from the Saqqez shear zone belt by leveraging available multielement geochemical information. This study introduces a hybrid correction framework based on a machine learning algorithm and targeted Winsorization (MLA–TW) that integrates TW with RF regression to reconstruct a realistic and geochemically plausible value for a highly influential Au outlier. Three scenarios were examined: (1) modeling with the original dataset containing a 739 ppb outlier, (2) modeling after removing the outlier, and (3) modeling with a reconstructed value obtained from the MLA–TW approach. The RF model showed reliable predictive capacity (R² = 0.85), and the reconstructed value preserved both geological plausibility and nonlinear multivariate structure. Application of the C–A fractal model demonstrated that the MLA–TW scenario yielded the most stable population breaks, the most robust anomaly thresholds, and the highest spatial fidelity, successfully identifying verified gold prospects and deposits in the region. Overall, the MLA–TW framework stabilizes the C–A model and improves its robustness by reducing the statistical leverage of extreme values while preserving the nonlinear geochemical patterns essential for anomaly detection. The results confirm that this intelligent hybrid approach provides an objective and geologically meaningful methodology for refining Au threshold determination and delineating shear zone-related gold targets with improved accuracy. Full article

The rapid advancement of autonomous driving technologies makes the effective coordination of vehicle groups a critical requirement for ensuring both safety and efficiency in smart urban environments. Although individual autonomous vehicles may operate correctly in isolation, their collective behavior can still lead to emergent issues such as deadlocks or collisions arising from complex inter-vehicle interactions. To address this challenge, we propose a hybrid automaton-based control framework for autonomous vehicle groups that integrates both normal and emergency operational modes to jointly guarantee safety and performance. In this paper, we present the formal specification and verification of the proposed system using rewriting logic and the Maude tool. Our main contributions are threefold: (1) the construction of detailed hybrid automata models that capture vehicle dynamics and decision-making; (2) the development of formal specifications in Maude from these models; and (3) the systematic verification of key system properties, including core safety invariants, such as collision avoidance, obstacle stopping, and velocity bounds. The verification results demonstrate that the proposed model consistently upholds safety conditions, ensures that vehicles come to a safe stop before encountering obstacles, and effectively prevents collisions within the group. Full article

Landmark recognition for smart tourism is usually validated on curated benchmark images. In deployment, however, the classifier must handle user-generated photographs whose viewpoint, lighting, resolution, occlusion, and compression differ sharply from curated data. This paper evaluates a previously published multi-threshold enhancement and selective YOLO11n-cls ensemble under this shift, and provides a preliminary zero-shot comparison of three general-purpose multimodal large language models (MLLMs) on the same task. To measure the shift, we build Samarkand v2-SNS, a 300-image out-of-distribution test set of social-media photographs of 12 Samarkand landmarks, disjoint from the training and validation data. Under the shift, four supervised baselines fall by 12.73–22.08 percentage points to 73–80% accuracy, and their in-distribution ranking does not hold. The selective ensemble degrades least (99.24% to 93.00%, −6.24 points) and outperforms the strongest baseline by 13 points. A capacity-matched ablation shows that most of this robustness comes from enhancement diversity, not from generic ensembling. In a preliminary comparison, zero-shot MLLMs (GPT-5, Claude Sonnet 4.5, Gemini 2.5) reach only 24.81–54.26%, far below deployment needs. The results argue for reporting out-of-distribution accuracy alongside curated benchmarks, and for hybrid systems that pair compact specialised recognisers with MLLM-based interpretation. Full article