Search Results (30,854)

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

Background/Objectives: Periodontitis is a highly prevalent inflammatory disease and a major cause of tooth loss worldwide. Accurate diagnosis requires integration of clinical and radiographic findings, but interpretation of panoramic radiographs is subject to variability. Artificial intelligence (AI) has emerged as a promising adjunct for radiographic assessment. This systematic review evaluated the diagnostic performance of AI-based models for detecting periodontitis using panoramic radiographic images. Methods: A systematic search of PubMed, Scopus, and Web of Science identified studies published between 1 January 2015 and 1 March 2026. Eligible studies assessed AI models for periodontitis detection on panoramic radiographs and used either clinically confirmed periodontal diagnosis or expert radiographic annotation as the reference standard. Data extraction and quality assessment were performed independently by two reviewers using the QUADAS-2 tool. Owing to heterogeneity in AI architectures, datasets, and outcome measures, a narrative synthesis was conducted. Results: Nine studies met the inclusion criteria, comprising more than 20,000 radiographs. AI models included convolutional neural networks (CNNs), segmentation-based systems, and hybrid architectures. Sensitivity ranged from 0.795 to 1.00, specificity from 0.784 to 0.99, and AUC values from 0.843 to 0.967. Studies using clinical periodontal diagnosis as the reference standard generally reported lower performance than those relying solely on expert annotation. Only four studies performed external validation, and dataset sizes varied widely. One study combining panoramic and periapical radiographs showed moderate diagnostic performance. Conclusions: AI-based diagnostic models demonstrate promising performance for detecting periodontitis on panoramic radiographs, with several studies reporting high sensitivity and AUC values. However, heterogeneity in reference standards, limited external validation, and inconsistent dataset quality restrict generalizability. AI should be considered an adjunct to, rather than a replacement for, comprehensive clinical periodontal examination. Standardized datasets and robust external validation are needed to support clinical implementation. Full article

In this paper, we propose a novel generative adversarial network (GAN)-based joint localization and mapping (JLAM) method using angle difference of arrival (ADOA) measurements for millimeter-wave (mmWave) communication systems. The proposed method adopts a deep auto-encoder neural network as the discriminator of the GAN and models the generator as an explicit geometric ADOA function of the access point (AP) positions and the mobile terminal (MT) position, rather than as a conventional black-box neural network. By exploiting the two-dimensional distribution characteristics of high-dimensional ADOA vectors collected at a large number of random and unknown MT positions, the proposed method learns the ADOA data distribution and transforms it into the AP geometric topology. Then, the MT positions and the indoor map are estimated based on the recovered physical and virtual AP topology. The simulation results show that, under the representative setting with $N=2000$ measured ADOA vectors and $\sigma =2^{\circ }$ AOA measurement noise, the proposed method achieves an average localization error of about 0.25 m, compared with about 0.60 m for the JADE algorithm, corresponding to an error reduction of approximately $58\%$. The proposed method also provides more accurate room boundary estimation than JADE, confirming its effectiveness for mmWave JLAM. Full article

To address the deep spatiotemporal coupling of source-load dual uncertainties attributed to the high penetration of distributed photovoltaics (PVs) and electric vehicles (EVs) into distribution grids, and the difficulty of analyzing composite disturbances using traditional methods, this paper proposes a voltage quality analysis method that considers spatiotemporal coupling of source-load uncertainty, focusing on steady-state voltage deviation and three-phase imbalance problems. First, a probabilistic model of PV generation is constructed using the beta distribution combined with Monte Carlo-based scenario reduction, and high-precision forecasting of EV charging loads is achieved by an attention-based convolutional neural network and long short-term memory network. Second, multi-scenario spatiotemporal power flow calculations are conducted on the distribution network to analyze the complementary effects of voltage deviation and three-phase imbalance under the hybrid integration of PV and EV. Finally, gray wolf optimization-based variational mode decomposition is introduced to adaptively decompose the source-load power. This reveals the intrinsic mechanisms where low-frequency components dominate the fundamental amplitude variations of bus voltages, while high-frequency components exert a significant impact on voltage quality. Simulation results demonstrate that the proposed method can effectively analyze the spatiotemporal coupling of source-load uncertainties, providing technical support for the comprehensive management of voltage quality in distribution networks. Full article

The trans-Neptunian region (30–50 AU) is a dynamically structured reservoir of icy planetesimals whose orbital architecture reflects resonant dynamics, chaotic transport, and long-term gravitational sculpting by the giant planets. This review synthesizes recent developments in the dynamical investigation of trans-Neptunian objects (TNOs), with an emphasis on mean-motion and secular resonances, as well as chaotic diffusion, in a system whose growing observational census makes it an ideal testbed for chaos detection methods. Classical indicators, including Lyapunov exponents, MEGNO, SALI/GALI, and frequency map analysis, provide the quantitative backbone for mapping TNO phase space and are complemented by modern approaches such as Lagrangian descriptors, the FAIR resonance identification method, entropy-based chaos indicators, and recurrence plot divergence methods. An anomalous diffusion framework, in which mean squared displacement scales as a power law in time, further enables classification of sub- and superdiffusive orbital transport. Machine learning has emerged as a powerful complement to traditional dynamical methods: surrogate classifiers, deep neural network solvers, and hybrid physics–data-driven frameworks together extend reliable prediction horizons in chaotic regimes and open new routes for Bayesian inference of migration scenarios. The review concludes that the most promising path forward lies in hybrid dynamical–statistical frameworks anchored to Hamiltonian dynamics, enabling efficient exploration of high-dimensional parameter spaces informed by the expanding body of trans-Neptunian observations. Full article

The safe disposal of spent fuel and high-level radioactive waste has become a critical bottleneck restricting the sustainable development of nuclear energy, and 3D geological modeling serves as a core technology for repository siting and safety assessment. Taking the upper member of the Lower Cretaceous Bayingobi Formation in the Tamusu area as the research object, this study focuses on sedimentary facies identification, lithofacies prediction, 3D geological modeling, and candidate site optimization. A convolutional neural network (CNN) + attention algorithm is proposed for high-precision lithofacies identification, and a Geo-CVAE-GAN model is constructed to address data sparsity and reconstruct 3D geological models. Following the workflow of single-well fine analysis, multi-method fusion prediction, and 3D geological modeling, the Sequential Indicator Simulation (SIS) algorithm is improved to build a 3D lithofacies model, and four-property parameter modeling is completed under facies control. Optimal sites are delineated via 3D spatial superimposition based on parameter thresholds. The results show that favorable mudstone layers display a dual-layer structure: stable thick layers in deep strata and thin superimposed layers in shallow strata. A preliminary total area of approximately 165 km² is identified in Preselected Sections I and II, with target intervals at a 400–800 m depth, mud content exceeding 75%, and excellent physical properties, including low porosity, low permeability, and low water saturation. This study reveals the spatial distribution of favorable mudstone in the Tamusu area, and the preferred zones fully meet the siting criteria for high-level radioactive waste repositories, providing a reliable geological basis and technical support for subsequent exploration and engineering design. Full article

Background/Objectives: Manual Pap smear analysis for cervical cancer screening is limited by inter-observer variability, time constraints, and restricted expert availability. Although convolutional neural networks (CNNs) have automated cervical cell classification, they remain limited in modeling long-range spatial dependencies and often lack clinical interpretability. Methods: In this study, Vision Transformer (ViT) architectures were systematically optimized to enhance automated cervical cancer screening and improve interpretability. The Herlev dataset (917 images: 242 normal, 675 abnormal) was utilized to optimize ViT-Tiny, a lightweight ViT architecture designed for reduced computational complexity, through a comprehensive evaluation of augmentation strategies, class weighting, and hyperparameters. Results: The optimal configuration achieved a cross-validation accuracy of approximately 95% (94.89% for the best replicated configuration), in which random horizontal flipping and class weighting (0.7 × 1.3) were identified as most effective. Gradient-weighted Class Activation Mapping (Grad-CAM) analysis confirmed that model attention corresponded to clinically relevant morphological features, including nuclei regions, cell boundaries, and chromatin texture, which align with cytopathological criteria. Conclusions: These findings indicate that Vision Transformers can deliver accurate and interpretable decision support for cervical cancer screening by combining competitive classification performance with attention-based transparency relevant to medical AI. Further validation on larger, multi-center datasets remains necessary before clinical deployment. Full article

Background/Objectives: Impacted third molars may adversely affect adjacent second molars, leading to pathological conditions such as external root resorption and dental caries. Accurate assessment of these interactions is important for treatment planning and clinical decision-making. Although cone-beam computed tomography (CBCT) provides detailed three-dimensional imaging, image interpretation remains challenging. Recent advances in artificial intelligence have enabled automated radiographic analysis using deep learning methods. Methods: This retrospective study included 162 CBCT scans obtained from patients aged 18–75 years. A total of 306 third molar–second molar units were evaluated. Based on radiographic findings, interactions were categorized as independent, contact, or resorption. Several deep learning architectures were developed and evaluated, including conventional convolutional neural networks (CNNs), attention-based CNNs, and Vision Transformer (ViT) models. Performance was assessed using standard classification metrics, and an ensemble approach was applied to improve predictive stability. Results: Attention-based and Transformer-based models generally outperformed conventional CNN architectures. These models achieved better discrimination among the defined classes and demonstrated superior overall performance. The ensemble model produced the most reliable results, achieving the highest macro-area under the curve (macro-AUC) values. Distinguishing contact cases from independent cases was the most challenging task, whereas resorption cases were identified more consistently across different models. Conclusions: Transformer-based deep learning models showed promising performance for CBCT-based assessment of third molar–second molar interactions. Ensemble learning further improved classification reliability and robustness. These findings suggest that artificial intelligence-assisted systems may support early detection of third molar-related pathological changes and contribute to more accurate radiological evaluation and clinical decision-making. Full article

Aluminum alloy hot rolling features diverse varieties, large variations in incoming strip thickness, and strong process nonlinearity. Traditional rolling force prediction models rely on simplified physical assumptions and poor adaptability, making it hard to satisfy high-precision production requirements. This paper presents a mechanism–data dual-driven PSO-BP neural network method for rolling force prediction which is applicable to the rolling temperature range of 320 °C to 520 °C. The SIMS mechanism model is employed as a physical constraint, and a hybrid PSO-GD algorithm optimizes the initial weights and thresholds of the BP network, avoiding the local optimum issue of conventional BP. The rolling mechanism model is embedded into the loss function to deeply integrate physical laws and data-driven learning. Validation using 508 sets of field data from 5083 aluminum alloy hot rolling shows that the model achieves a M_APE of 5.0794% and R² of 0.9254, significantly outperforming the traditional mechanism model (8.91%) and standard BP (8.77%). The proposed model preserves physical interpretability while utilizing data-driven adaptability, offering an effective approach for high-precision rolling force prediction and improving the dimensional accuracy of hot-rolled aluminum alloy sheets. Full article

Background/Objectives: Accurate estimation of meal composition from food images can support safer and more reliable insulin bolus decision-making for individuals with Type 1 diabetes. Existing food recognition and nutrition estimation systems are often designed for general dietary logging and do not directly integrate food analysis with personalized insulin therapy parameters. Methods: This study presents an image-based nutrition estimation and insulin decision-support module developed within the AI-assisted Diabetes Care (AIDCARE) platform. The proposed system uses a convolutional neural network (CNN) to classify food items from a single meal image, and retrieves reference nutritional values from a food composition database. A separate multimodal large language model (MLLM)-based estimation component is then used to estimate portion size, allowing carbohydrate and nutrient values to be scaled according to the observed serving. Results: A curated food image dataset containing 40 food categories was used to evaluate three CNN architectures: ResNet50, Inception V3, and EfficientNet-B0. EfficientNet-B0 achieved the best classification performance, with 94.91% validation accuracy, 95.55% precision, 94.87% recall, and 94.90% F1-score. The portion-estimation component achieved an MAE of 12.27 g and an RMSE of 15.11 g. The estimated carbohydrate value is combined with user-specific clinical parameters, including the insulin-to-carbohydrate ratio and insulin sensitivity factor, to generate advisory bolus guidance. To support safety, the system requires user confirmation or correction of the recognized food category and estimated portion before insulin guidance is displayed. Conclusions: The proposed system is intended for advisory decision support only and is not designed to replace clinical judgment or autonomous insulin delivery systems. Full article

Background/Objectives: Gastrointestinal bleeding is a critical finding in wireless capsule endoscopy (WCE), but manual examination of thousands of image frames is labor-intensive, time-consuming, and susceptible to missed lesions. This study aimed to develop and evaluate a comprehensive deep-learning framework for automated bleeding detection, localization, and quantitative assessment in WCE images. Methods: The proposed framework integrates three complementary deep-learning models: (i) a custom two-dimensional convolutional neural network (2D-CNN) for frame-level bleeding classification, (ii) a three-dimensional convolutional neural network (3D-CNN) for sequence-level analysis by exploiting temporal information from consecutive frames, and (iii) a U-Net architecture for pixel-level segmentation and bleeding-area quantification. The models were trained and evaluated using expert-annotated WCE datasets with pixel-level ground-truth masks. Results: The proposed 2D-CNN and 3D-CNN achieved excellent classification performance, with areas under the receiver operating characteristic curve (AUCs) of 0.9986 and 0.9971, respectively. The U-Net model achieved a Dice similarity coefficient of 0.93, an intersection-over-union (IoU) of 0.8677, and an overall segmentation accuracy of 97.25%. The integrated framework outperformed previously reported methods, demonstrating robust performance for bleeding detection, localization, and quantitative assessment. Conclusions: The proposed end-to-end deep-learning framework enables accurate automated bleeding detection, localization, and severity quantification in WCE images. By reducing the burden of manual image review, improving diagnostic consistency, and providing objective bleeding assessment, the framework has strong potential to support clinical decision-making and enhance gastrointestinal diagnostic workflows. Full article

Accurate and robust localization is essential for intelligent vessels operating in complex marine and port environments. However, single-sensor localization is often affected by limited observation range, environmental occlusion, local interference, and sensor degradation. Although multi-sensor fusion can improve localization reliability, unknown cross-correlated measurement noise arising from shared disturbances, time synchronization errors, communication delays, and inconsistent fusion rates may degrade traditional information-filter-based fusion methods. To address this problem, this paper proposes an Adaptive Correlation Information Filter Network (ACIFNet) for multi-sensor fusion localization of intelligent vessels. ACIFNet preserves the recursive structure of the extended information filter and uses a Transformer-based network to learn adaptive information-domain fusion weights, thereby compensating for unknown inter-sensor correlations without explicitly estimating the full correlation covariance matrix. Experiments on constant-velocity, coordinated-turn (CV), and three-degree-of-freedom vessel motion models, together with a real-world restricted-waterway dataset, demonstrate that ACIFNet achieves higher localization accuracy and stability than Edge Incorporative Fusion (EIF)-inexact fusion, measurement fusion, and KalmanNet. In the CV and three-degree-of-freedom experiments, ACIFNet reduces the mean RMSE by 48.7%, 23.2%, and 26.1%, respectively, compared with KalmanNet. On the real-world dataset, ACIFNet achieves a mean position error of 9.90 m, an RMSE of 11.24 m, and a cross-track error of 8.72 m. These results show that ACIFNet effectively combines the interpretability of information filtering with the adaptive representation capability of neural networks for robust multi-sensor fusion localization under unknown cross-correlated measurement noises. Full article

Introduction: Deep learning has emerged as a promising approach for automated bone age assessment; however, the scientific development of this field has not been comprehensively characterized. Materials and Methods: A bibliometric analysis of studies published between 2015 and 2025 was conducted using data from Web of Science, Scopus, MEDLINE (PubMed), IEEE Xplore, and Google Scholar. A total of 67 studies were included. Bibliometric mapping, network visualization, and knowledge structure analyses were performed using VOSviewer^® (v. 1.6.20) and CiteSpace^® (v. 6.3.R1). Results: Scientific output increased markedly, from 1 publication in 2015–2016 to 25 in 2025. The period 2019–2020 achieved the highest citation impact (1191 citations across 18 studies). China was the leading contributor, followed by the United States, South Korea, and Turkey. Keyword analysis identified convolutional neural networks, the RSNA dataset, and the Greulich–Pyle method as central research themes. Recent studies have increasingly focused on external validation, multicenter datasets, and transformer-based architectures. Conclusions: Research on deep learning for bone age assessment has grown substantially over the past decade. This bibliometric analysis highlights the field’s major contributors, influential publications, and emerging trends, while emphasizing the need for greater validation, dataset diversity, and standardized reporting. Full article

Regression models are widely used for continuous function approximation in applied research, yet selecting an appropriate model remains challenging for applied users who must balance predictive accuracy, interpretability, robustness, computational cost, and preprocessing requirements. This methodological review provides a decision-oriented synthesis of regression model families, preprocessing strategies, and evaluation criteria for transparent and reproducible model selection. The reviewed methods are organized by modeling principle, including linear and regularized models, robust and distribution-aware estimators, online learning methods, tree-based ensembles, kernel-based and probabilistic approaches, instance-based regressors, neural networks, and symbolic regression. The main contribution is a practical framework that connects data characteristics, including linearity, dimensionality, feature scale, target distribution, noise, outliers, and sample size, with suitable model families, preprocessing choices, and performance metrics. The review distinguishes theoretical guarantees, empirical tendencies, and implementation-dependent behavior because properties such as robustness, interpretability, scalability, and approximation capacity cannot be reduced to universal binary categories. The resulting comparative tables and decision criteria provide a compact reference for applied researchers designing regression workflows that are theoretically grounded, practically feasible, and aligned with research objectives. Full article

In social sensing environments, misinformation and coordinated manipulation often manifest through implicit semantic signals, complex behavioral dynamics, and highly coupled propagation structures. These factors pose significant challenges to artificial intelligence-driven sensing systems regarding data quality, multimodal fusion, and robustness. To address these issues, this study proposes an artificial intelligence-driven multi-granularity sensing framework. This framework integrates heterogeneous sensing signals from post-level semantic perception, user-level behavioral sensing, and group-level structural sensing into a unified representation space. Hierarchical consistency constraints enable cross-granularity sensing collaboration. This mechanism enhances stability and discriminative capability under complex and noisy data conditions. Methodologically, the framework jointly incorporates semantic sensing via text encoding, temporal sensing via behavioral sequence modeling, and structural sensing via graph neural network-based propagation. This integration effectively mitigates information bias induced by single-perspective sensing and improves the modeling of latent risk patterns. Experimental results on real-world datasets demonstrate that the proposed framework achieves significant improvements across multiple evaluation metrics. Specifically, it achieves a Precision of 0.847, a Recall of 0.812, an F1-score of 0.829, an Accuracy of 0.856, and an Area Under Curve of 0.913. It consistently outperforms traditional machine learning models, as well as mainstream deep learning and graph-based approaches. Furthermore, comparison experiments validate the complementarity among semantic, behavioral, and structural sensing signals. The full model achieves an improvement of more than 3 percentage points in the F1-score compared to single-granularity configurations. An ablation study further demonstrates that each sensing module contributes substantially to performance enhancement, with the semantic sensing and hierarchical consistency constraints playing particularly critical roles. Overall, the proposed method exhibits a strong capability to handle complex heterogeneous sensing data. It improves robustness and enhances cross-level information utilization, providing an effective solution to data-related challenges in artificial intelligence-driven sensing systems. Full article