Search Results (3,693)

Quaternion-valued neural networks (QVNNs) that have multiple types of delays (leakage, time-varying, distributed, and neutral) and defined on time scales are discussed in this paper. Quaternions form a 4D normed division algebra and allow for a better representation of 3D and 4D data. QVNNs have been proposed and applications have appeared lately. Time-scale calculus was developed to allow the joint treatment of systems, or any hybrid mixing of them, and was also applied with success to the analysis of dynamic properties for neural networks (NNs). Because of its generality, encompassing the common properties of discrete-time (DT) and continuous-time (CT) NNs, time-scale NNs dynamics research does not benefit from a fully-developed Lyapunov theory. So, Halanay-type inequalities have to be used instead. To this end, we provide a novel generalization of inequalities of Halanay-type on time scales specifically suited for neutral systems, i.e., systems with neutral delays. Then, this new lemma is employed to obtain sufficient conditions presented both as linear matrix inequalities (LMIs) and as algebraic inequalities for the exponential stability and exponential synchronization of QVNNs on time scales with the mentioned delay types. The model put forward in this paper has a generality which is appealing for practical applications, in which both DT and CT dynamics are interesting, and all the discussed types of delays appear. For both the DT and CT scenarios, four numerical applications are used to illustrate the four theorems put forward in this research. Full article

Existing 3D point cloud enhancement methods typically rely on artificially designed geometric transformations or local blending strategies, which are prone to introducing illogical deformations, struggle to preserve global structure, and exhibit insufficient adaptability to diverse degradation patterns. To address these limitations, this paper proposes SFD-ADNet—an adaptive deformation framework based on a dual spatial–frequency domain. It achieves 3D point cloud augmentation by explicitly learning deformation parameters rather than applying predefined perturbations. By jointly modeling spatial structural dependencies and spectral features, SFD-ADNet generates augmented samples that are both structurally aware and task-relevant. In the spatial domain, a hierarchical sequence encoder coupled with a bidirectional Mamba-based deformation predictor captures long-range geometric dependencies and local structural variations, enabling adaptive position-aware deformation control. In the frequency domain, a multi-scale dual-channel mechanism based on adaptive Chebyshev polynomials separates low-frequency structural components from high-frequency details, allowing the model to suppress noise-sensitive distortions while preserving the global geometric skeleton. The two deformation predictions dynamically fuse to balance structural fidelity and sample diversity. Extensive experiments conducted on ModelNet40-C and ScanObjectNN-C involved synthetic CAD models and real-world scanned point clouds under diverse perturbation conditions. SFD-ADNet, as a universal augmentation module, reduces the mCE metrics of PointNet++ and different backbone networks by over 20%. Experiments demonstrate that SFD-ADNet achieves state-of-the-art robustness while preserving critical geometric structures. Furthermore, models enhanced by SFD-ADNet demonstrate consistently improved robustness against diverse point cloud attacks, validating the efficacy of adaptive space-frequency deformation in robust point cloud learning. Full article

Similarity search on image embeddings is a common practice for image retrieval in machine learning and pattern recognition systems. Approximate nearest neighbor (ANN) methods enable scalable similarity search on large datasets, often approaching sub-linear complexity. Yet, little empirical work has examined how ANN neighborhood geometry differs from that of exact k-nearest neighbors (k-NN) search as the neighborhood size increases under constrained search effort. This study quantifies how approximate neighborhood structure changes relative to exact k-NN search as k increases across three experimental conditions. Using multiple random subsets of 10,000 images drawn from the STL-10 dataset, we compute ResNet-50 image embeddings, perform an exact k-NN search, and compare it to a Hierarchical Navigable Small World (HNSW)-based ANN search under controlled hyperparameter regimes. We evaluated the fidelity of neighborhood structure using neighborhood overlap, average neighbor distance, normalized barycenter shift, and local intrinsic dimensionality (LID). Results show that exact k-NN and ANN search behave nearly identically when $efSearch>k$. However, as the neighborhood size grows and $efSearch$ remains fixed, ANN search fails abruptly, exhibiting extreme divergence in neighbor distances at approximately $k\approx 2$–$3.5\times efSearch$. Increasing index construction quality delays this failure, and scaling search effort proportionally with neighborhood size ($efSearch=\alpha \times k$ with $\alpha \ge 1$) preserves neighborhood geometry across all evaluated metrics, including LID. The findings indicate that ANN search preserves neighborhood geometry within its operational capacity but abruptly fails when this capacity is exceeded. Documenting this behavior is relevant for scientific applications that approximate embedding spaces and provides practical guidance on when ANN search is interchangeable with exact k-NN and when geometric differences become nontrivial. Full article

An adaptive control strategy is developed and analyzed for trajectory tracking of mechanical systems subject to simultaneous model uncertainties and full-state constraints. To overcome the significant hurdle of guaranteeing both transient and steady-state performance within a user-defined time, a novel predefined-time adaptive neural network (NN) control scheme is proposed. By integrating predefined-time stability theory with a nonlinear mapping framework, a control scheme is developed to rigorously enforce full-state constraints while achieving predefined-time convergence. Radial basis function neural networks (RBFNNs) are employed to approximate the unknown system dynamics, with adaptive laws designed for online learning. The nonlinear mapping is strategically incorporated to ensure that the full-state constraints are never violated throughout the entire operation. Furthermore, through Lyapunov stability theory, it is proved that all signals of the resulting closed-loop system are uniformly ultimately bounded, and most importantly, the trajectory tracking error converges to a small neighborhood of zero within a predefined time, which can be explicitly set regardless of initial conditions. Comparative simulation results on a representative mechanical system are provided to demonstrate the superiority of the proposed controller, showcasing its faster convergence, higher tracking accuracy, and guaranteed constraint satisfaction compared to conventional finite-time and adaptive NN control methods. Full article

Artificial intelligence (AI) has revolutionized medical image analysis. Several neural network (NN) architectures were developed and applied across the last decade, becoming essential for automated diagnosis and clinical applications. AI based on NNs has become increasingly integrated into gastroenterology, offering new opportunities for automated lesion detection and workflow optimization. Small-bowel capsule endoscopy (SBCE) has benefited substantially from these advances, addressing long-standing challenges such as time-consuming video review and variability among readers. This systematic review and meta-analysis evaluated neural network-based models for lesion detection in SBCE, assessing pooled diagnostic accuracy and the impact of AI on reading time. A total of 44 primary studies were included: 36 validation studies for accuracy and 9 clinical studies for reading time. All NN architectures demonstrated high diagnostic performance, with a pooled accuracy of 95.3% (95% CI: 94.1–96.5%). More recent architectures, including transformer-based and capsule networks, outperformed classical convolutional neural networks (CNNs). AI assistance significantly reduced SBCE reading time, with a pooled mean reduction of 84% compared to standard review. These findings highlight the strong potential of AI to enhance SBCE efficiency and diagnostic reliability. Full article

This study develops a comprehensive workflow integrating Headspace Solid-Phase Microextraction Gas Chromatography–Mass Spectrometry (HS-SPME-GC-MS) with advanced supervised machine learning to authenticate the botanical origin of honeys from five distinct floral sources—coriander, orange blossom, astragalus, rosemary, and chehelgiah. While HS-SPME-GC-MS combined with traditional chemometrics (e.g., PCA, LDA, OPLS-DA) is well-established for honey discrimination, the application and direct comparison of Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Neural Network (NN) models represent a significant advancement in multiclass prediction accuracy and model robustness. A total of 57 honey samples were analyzed to generate detailed volatile organic compound (VOC) profiles. Key chemotaxonomic markers were identified: anethole in coriander and chehelgiah, thymoquinone in astragalus, p-menth-8-en-1-ol in orange blossom, and dill ester (3,6-dimethyl-2,3,3a,4,5,7a-hexahydrobenzofuran) in rosemary. Principal component analysis (PCA) revealed clear separation across botanical classes (PC1: 49.8%; PC2: 22.6%). Three classification models—RF, XGBoost, and NN—were trained on standardized, stratified data. The NN model achieved the highest accuracy (90.32%), followed by XGBoost (86.69%) and RF (83.47%), with superior per-class F1-scores and near-perfect specificity (>0.95). Confusion matrices confirmed minimal misclassification, particularly in the NN model. This work establishes HS-SPME-GC-MS coupled with deep learning as a rapid, sensitive, and reliable tool for multiclass honey botanical authentication, offering strong potential for real-time quality control, fraud detection, and premium market certification. Full article

Within the traditional electronic neural network framework, Generative Adversarial Networks (GANs) have achieved extensive applications across multiple domains, including image synthesis, style transfer and data augmentation. Recently, several studies have explored the use of optical neural networks represented by the diffractive deep neural network (D²NN) for GANs. However, most of these focus on applications of the generative network, and there is currently no well-established D²NN architecture that simultaneously implements generative adversarial functionality. Here, we propose a novel implementation scheme for generative adversarial networks based on all-optical diffraction layers, demonstrating a complete all-optical adversarial architecture that simultaneously realizes both the generative network and the adversarial network (D²NN-GAN). We validated this method on the MNIST handwritten digit dataset, achieving Nash equilibrium convergence with the discriminator accuracy stabilizing around 50%. Concurrently, the average SSIM parameter of generated images reached 0.9573, indicating that the generated samples possess high quality and closely resemble real samples. Furthermore, we extended the framework to the KTH human action dataset, successfully reconstructing the “running” action with a discriminator accuracy of approximately 75%. The D²NN-GAN architecture introduces a fully optical generative adversarial model, providing a practical path for future optical modeling methods, such as image generation and video synthesis. Full article

Glucocorticoids are widely applied intra-articularly to alleviate inflammation and pain in osteoarthritis (OA). However, repeated administration and high local concentrations can lead to crystal deposition on the cartilage surface, contributing to chondrocyte damage and extracellular matrix (ECM) degradation, potentially accelerating OA progression. Calcium-dependent mechanosensors play a critical role in mediating catabolic responses in chondrocytes, but it remains unclear whether glucocorticoids affect chondrocyte mechanosensitivity or biomechanical properties. This in vitro study examined the dose-dependent effects of triamcinolone acetonide (TA) on chondrocyte biomechanics and mechanosensitivity. Primary human chondrocytes (N = 23) were cultured for one week with TA (2 µM–2 mM) or control medium. Cytoskeletal organization was visualized by F-actin staining (N = 6), and cellular elasticity (N = 5) was quantified via atomic force microscopy (AFM). Mechanotransduction was analyzed by Ca²⁺ imaging (Fluo-4 AM) upon AFM-based indentation (500 nN). Expression of matrix-related and mechanosensitive genes (N = 9) was assessed by qPCR. TA exposure induced a concentration-dependent reorganization of the F-actin cytoskeleton, pronounced at 0.2 mM, accompanied by a significant increase in the elastic modulus (p < 0.001). TA further augmented Ca²⁺ fluorescence intensity under basal conditions and during mechanical stimulation. Blocking cationic mechanosensitive channels with GsMtx4 (N = 3) markedly reduced the TA-evoked Ca²⁺ influx (p < 0.0001). Significant reduction in MMP1 was observed on the transcriptional level (N = 9) after TA-treatment (p < 0.05). In summary, TA enhances chondrocyte stiffness through cytoskeletal condensation and amplifies Ca²⁺-dependent mechanotransduction but reduces MMP1 expression, indicating a dual biomechanical response of chondrocytes to OA under exposure of potent corticosteroid. Full article

This study contributes to the advancement of circular economy practices in polymer manufacturing by applying machine learning algorithms (MLA) to predict the tensile strength of recycled low-density polyethylene (LDPE) building films. As the construction and packaging industries increasingly seek eco-efficient and low-carbon materials, recycled LDPE offers a valuable route toward sustainable resource management. However, ensuring consistent mechanical performance remains a challenge when reusing polymer waste streams. To address this, tensile tests were conducted on LDPE films produced from recycled granules, measuring tensile strength, strain, mass per unit area, thickness, and surface roughness. Three established machine learning algorithms—feed-forward Neural Network (NN), Gradient Boosting Machine (GBM), and Extreme Gradient Boosting (XGBoost)—were implemented, trained, and optimized using the experimental dataset using R statistical software (version 4.4.3). The models achieved high predictive accuracy, with XGBoost providing the most robust performance and the highest level of explainability. Feature importance analysis revealed that mass per unit area and surface roughness have a significant influence on film durability and performance. These insights enable more efficient production planning, reduced raw material usage, and improved quality control, key pillars of sustainable technological innovation. The integration of data-driven methods into polymer recycling workflows demonstrates the potential of artificial intelligence to accelerate circular economy objectives by enhancing process optimization, material performance, and resource efficiency in the plastics sector. Full article

Objective: To develop and validate a genetic diagnostic model for colorectal cancer (CRC). Methods: First, differential expression genes (DEGs) between colorectal cancer and normal groups were screened using the TCGA database. Subsequently, a two-sample Mendelian randomization analysis was performed using the eQTL genomic data from the IEU OpenGWAS database and colorectal cancer outcomes from the R12 Finnish database to identify associated genes. The intersecting genes from both methods were selected for the development and validation of the CRC genetic diagnostic model using nine machine learning algorithms: Lasso Regression, XGBoost, Gradient Boosting Machine (GBM), Generalized Linear Model (GLM), Neural Network (NN), Support Vector Machine (SVM), k-Nearest Neighbors (KNN), Random Forest (RF), and Decision Tree (DT). Results: A total of 3716 DEGs were identified from the TCGA database, while 121 genes were associated with CRC based on the eQTL Mendelian randomization analysis. The intersection of these two methods yielded 27 genes. Among the nine machine learning methods, XGBoost achieved the highest AUC value of 0.990. The top five genes predicted by the XGBoost method—RIF1, GDPD5, DBNDD1, RCCD1, and CLDN5—along with the five most significantly differentially expressed genes (ASCL2, IFITM3, IFITM1, SMPDL3A, and SUCLG2) in the GSE87211 dataset, were selected for the construction of the final colorectal cancer (CRC) genetic diagnostic model. The ROC curve analysis revealed an AUC (95% CI) of 0.9875 (0.9737–0.9875) for the training set, and 0.9601 (0.9145–0.9601) for the validation set, indicating strong predictive performance of the model. SHAP model interpretation further identified IFITM1 and DBNDD1 as the most influential genes in the XGBoost model, with both making positive contributions to the model’s predictions. Conclusions: The gene expression profile in colorectal cancer is characterized by enhanced cell proliferation, elevated metabolic activity, and immune evasion. A genetic diagnostic model constructed based on ten genes (RIF1, GDPD5, DBNDD1, RCCD1, CLDN5, ASCL2, IFITM3, IFITM1, SMPDL3A, and SUCLG2) demonstrates strong predictive performance. This model holds significant potential for the early diagnosis and intervention of colorectal cancer, contributing to the implementation of third-tier prevention strategies. Full article

This study focuses on classifying electromyographic (EMG) signals to identify seven specific hand movements, including complete hand closure, individual finger closures, and a pincer grip. Accurately distinguishing these movements is challenging due to overlapping muscle activation patterns. To address this, a methodology structured in five stages was developed: placement of electrodes on specific forearm muscles to capture electrical activity during movements; acquisition of EMG signals from twelve participants performing the seven types of movements; preprocessing of the signals through filtering and rectification to enhance quality, followed by the extraction of features from three distinct types of preprocessed signals—filtered, rectified, and envelope signals—to facilitate analysis in the temporal, frequency, and time–frequency domains; extraction of relevant features such as amplitude, shape, symmetry, and frequency variance; and classification of the signals using eight machine learning algorithms: support vector machine (SVM), multiclass logistic regression, k-nearest neighbors (k-NN), Bayesian classifier, artificial neural network (ANN), random forest, XGBoost, and LightGBM. The performance of each algorithm was evaluated using different sets of features derived from the preprocessed signals to identify the most effective approach for classifying hand movements. Additionally, the impact of various signal representations on classification accuracy was examined. Experimental results indicated that some algorithms, especially when an expanded set of features was utilized, achieved improved accuracy in classifying hand movements. These findings contribute to the development of more efficient control systems for myoelectric prostheses and offer insights for future research in EMG signal processing and pattern recognition. Full article

Background: The persistently low 5-year survival rate for pancreatic cancer (PC) underscores the critical need for early detection. However, population-wide screening remains impractical. Artificial Intelligence (AI) models using electronic health record (EHR) data offer a promising avenue for pre-symptomatic risk stratification. Objective: To systematically review and meta-analyze the performance of AI models for PC prediction based exclusively on structured EHR data. Methods: We systematically searched PubMed, MedRxiv, BioRxiv, and Google Scholar (2010–2025). Inclusion criteria encompassed studies using EHR-derived data (excluding imaging/genomics), applying AI for PC prediction, reporting AUC, and including a non-cancer cohort. Two reviewers independently extracted data. Random-effects meta-analysis was performed for AUC, sensitivity (Se), and specificity (Sp) using R software version 4.5.1. Heterogeneity was assessed using I² statistics and publication bias was evaluated. Results: Of 946 screened records, 19 studies met the inclusion criteria. The pooled AUC across all models was 0.785 (95% CI: 0.759–0.810), indicating good overall discriminatory ability. Neural Network (NN) models demonstrated a statistically significantly higher pooled AUC (0.826) compared to Logistic Regression (LogReg, 0.799), Random Forests (RF, 0.762), and XGBoost (XGB, 0.779) (all p < 0.001). In analyses with sufficient data, models like Light Gradient Boosting (LGB) showed superior Se and Sp (99% and 98.7%, respectively) compared to NNs and LogReg, though based on limited studies. Meta-analysis of Se and Sp revealed extreme heterogeneity (I² ≥ 99.9%), and the positive predictive values (PPVs) reported across studies were consistently low (often < 1%), reflecting the challenge of screening a low-prevalence disease. Conclusions: AI models using EHR data show significant promise for early PC detection, with NNs achieving the highest pooled AUC. However, high heterogeneity and typically low PPV highlight the need for standardized methodologies and a targeted risk-stratification approach rather than general population screening. Future prospective validation and integration into clinical decision-support systems are essential. Full article

Background/Objectives: Arboviral diseases share common vectors, geographic distribution, and symptoms. Developing Machine Learning diagnostic tools for co-circulating arboviral diseases faces data-scarcity challenges. This study aimed to demonstrate that proof of concept using synthetic data can establish computational feasibility and guide future real-world validation efforts. Methods: We assembled a synthetic dataset of 28,000 records, with 7000 for each disease—Dengue, Zika, and Chikungunya—plus Influenza as a negative control. These records were obtained from the existing literature. A binary matrix with 67 symptoms was created for detailed statistical analysis using Odds Ratios, Chi-Square, and symptom-specific conditional prevalence to validate the clinical relevance of the simulated data. This dataset was used to train and evaluate various algorithms, including Multi-Layer Perceptron (MLP), Narrow Neural Network (NN), Quadratic Support Vector Machine (QSVM), and Bagged Tree (BT), employing multiple performance metrics: accuracy, precision, sensitivity, specificity, F1-score, AUC-ROC, and Cohen’s kappa coefficient. Results: The dataset aligns with the PAHO guidelines. Similar findings are observed in other arboviral databases, confirming the validity of the synthetic dataset. A notable performance across all evaluated metrics was observed. The NN model achieved an overall accuracy of 0.92 and an AUC above 0.98, with precision, sensitivity, and specificity values exceeding 0.85, and an average Uniform Cohen’s Kappa of 0.89, highlighting its ability to reliably distinguish between Dengue and Influenza, with a slight decrease between Zika and Chikungunya. Conclusions: These models could accelerate early diagnosis of arboviral diseases by leveraging encoded symptom features for Machine Learning and Deep Learning approaches, serving as a support tool in regions with limited healthcare access without replacing clinical medical expertise. Full article

Background and Objectives: Coronary artery disease (CAD) is a leading cause of morbidity and mortality worldwide. An early and accurate diagnosis is essential for effective clinical management and risk stratification. Recent advances in machine learning (ML) have provided opportunities to enhance the diagnostic performance by integrating multidimensional patient data. This study aimed to develop and compare several supervised ML algorithms for early CAD diagnosis using demographic, anthropometric, biochemical, and psychosocial parameters. Materials and Methods: A total of 300 adult patients (165 CAD-positive and 135 controls) were retrospectively analyzed using a dataset comprising 21 biochemical markers, body composition metrics, and self-reported eating behavior scores. Six ML algorithms, k-nearest neighbors (k-NNs), support vector machines (SVMs), artificial neural networks (ANNs), logistic regression (LR), naïve Bayes (NB), and decision trees (DTs), were trained and evaluated using 10-fold cross-validation. Model performance was assessed based on accuracy, sensitivity, false-negative rate, and area under the Receiver Operating Characteristic (ROC) curve (AUC). Results: The k-NN model achieved the highest performance, with 98.33% accuracy and an AUC of 0.99, followed by SVM (96.67%, AUC = 0.95) and ANN (95.33%, AUC = 0.98). Patients with CAD exhibited significantly higher levels of glucose, triglycerides (TGs), LDL cholesterol (LDL-C), and abdominal obesity, while vitamin B12 levels were lower (p < 0.001). Although emotional and mindful eating scores differed significantly between the groups, their contribution to model performance was limited. Conclusions: Machine learning models, particularly k-NN, SVM, and ANN, have demonstrated high accuracy in distinguishing CAD patients from healthy controls when applied to a diverse set of clinical and behavioral variables. This study highlights the potential of integrating psychosocial and clinical data to enhance CAD prediction models beyond traditional biomarkers. Full article

Background/Objectives: The application of deep learning models for rare diseases faces significant difficulties due to severe data scarcity. The detection of focal hyperostosis (PAH) is a crucial radiological sign for the surgical planning of sinonasal inverted papilloma, yet data is often limited. This study introduces and validates a robust methodological framework for building clinically meaningful deep learning models under extremely limited data conditions (n = 20). Methods: We propose a few-shot learning framework based on the nnU-Net architecture, which integrates an in-domain transfer learning strategy (fine-tuning a pre-trained skull segmentation model) to address data scarcity. To further enhance robustness, a specialized data augmentation technique called “window shifting” is introduced to simulate inter-scanner variability. The entire framework was evaluated using a rigorous 5-fold cross-validation strategy. Results: Our proposed framework achieved a stable mean Dice Similarity Coefficient (DSC) of 0.48 ± 0.06. This performance significantly outperformed a baseline model trained from scratch, which failed to converge and yielded a clinically insignificant mean DSC of 0.09 ± 0.02. Conclusions: The analysis demonstrates that this methodological approach effectively overcomes instability and overfitting, generating reproducible and valuable predictions suitable for rare data types where large-scale data collection is not feasible. Full article