Consultar por División

FUNIBER (139)
- Universidad de La Romana (47)
  - Investigación (47)
    - Producción Científica (39)

Agrupar por: Fecha | Título | Autores | Tipo de Documento | Sin Agrupar

Ir a: Artículo

Número de registros en este nivel: 39.

Artículo

Advanced Wafer Hotspot Detection through Image Segmentation and Stacked Model.

Advancing fake news combating using machine learning: a hybrid model approach.

Association between blood cortisol levels and numerical rating scale in prehospital pain assessment.

With the growing academic pressure and competitive educational environment, students often face mental stress, which can affect their academic performance and mental health. Its accurate and timely detection and prevention is important. Traditionally, mental stress has been reported by self-assessment, which is highly subjective and can be erroneous. With advances in neuroscience, electroencephalogram (EEG) signals have been used to study brain states more objectively. EEG-based features, including time-domain, frequency-domain, and various types of connectivity features, have been used to effectively classify stress signals. However, these individual features are only able to present one aspect of the brain under stress. Several studies have combined a distinct set of features extracted from EEG signals, including time and frequency domain features, with other peripheral signals. Stress is a complex mechanism which leads to alternation in brain dynamics, its connectivity patterns and information flow. This study proposed a feature-fusion model that can effectively combine spatial features, i.e. Microstates (MS), connectivity features like Transfer Entropy (TE) and Granger Causality (GC), which provided a new neuromarker for stress classification. These features are combined with attention fusion, which enhances the discriminant features and mitigates the individual limitations within each modality. We also extracted microstates for stress-based signals. It provided a new set of microstate topomaps to study brain networks when under stress, which was not explored previously. The proposed Attention-fusion based multi-feature set is classified using Support Vector Machine, Linear Discriminant Analysis (LDA) and Multilayer Perceptron (MLP) and gave a reliable accuracy of 95.47%, 98.91%, and 83.49%, respectively. To validate the proposed method, the classification results were compared with individual and binary fusion of MS, TE and GC features, which further confirmed the robustness of the framework. This proposed feature fusion provides a more robust stress classification neuromarker, which can effectively cover the brain dynamics for accurate reporting of the underlying mental state.

Attention-based multi-feature fusion neuromarker for EEG-driven stress classification in learners.

Histopathological evaluation is necessary for the diagnosis and grading of prostate cancer, which is still one of the most common cancers in men globally. Traditional evaluation is time-consuming, prone to inter-observer variability, and challenging to scale. The clinical usefulness of current AI systems is limited by the need for comprehensive pixel-level annotations. The objective of this research is to develop and evaluate a large-scale benchmarking study on a weakly supervised deep learning framework that minimizes the need for annotation and ensures interpretability for automated prostate cancer diagnosis and International Society of Urological Pathology (ISUP) grading using whole slide images (WSIs). This study rigorously tested six cutting-edge multiple instance learning (MIL) architectures (CLAM-MB, CLAM-SB, ILRA-MIL, AC-MIL, AMD-MIL, WiKG-MIL), three feature encoders (ResNet50, CTransPath, UNI2), and four patch extraction techniques (varying sizes and overlap) using the PANDA dataset (10,616 WSIs), yielding 72 experimental configurations. The methodology used distributed cloud computing to process over 31 million tissue patches, implementing advanced attention mechanisms to ensure clinical interpretability through Grad-CAM visualizations. The optimum configuration (UNI2 encoder with ILRA-MIL, 256 256 patches, 50% overlap) achieved 78.75% accuracy and 90.12% quadratic weighted kappa (QWK), outperforming traditional methods and approaching expert pathologist-level diagnostic capability. Overlapping smaller patches offered the best balance of spatial resolution and contextual information, while domain-specific foundation models performed noticeably better than generic encoders. This work is the first large-scale, comprehensive comparison of weekly supervised MIL methods for prostate cancer diagnosis and grading. The proposed approach has excellent clinical diagnostic performance, scalability, practical feasibility through cloud computing, and interpretability using visualization tools.

Benchmarking multiple instance learning architectures from patches to pathology for prostate cancer detection and grading using attention-based weak supervision.

Introduction: Jackfruit cultivation is highly affected by leaf diseases that reduce yield, fruit quality, and farmer income. Early diagnosis remains challenging due to the limitations of manual inspection and the lack of automated and scalable disease detection systems. Existing deep-learning approaches often suffer from limited generalization and high computational cost, restricting real-time field deployment. Methods: This study proposes CNNAttLSTM, a hybrid deep-learning architecture integrating Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) units, and an attention mechanism for multi-class classification of algal leaf spot, black spot, and healthy jackfruit leaves. Each image is divided into ordered 56×56 spatial patches, treated as pseudo-temporal sequences to enable the LSTM to capture contextual dependencies across different leaf regions. Spatial features are extracted via Conv2D, MaxPooling, and GlobalAveragePooling layers; temporal modeling is performed by LSTM units; and an attention mechanism assigns adaptive weights to emphasize disease-relevant regions. Experiments were conducted on a publicly available Kaggle dataset comprising 38,019 images, using predefined training, validation, and testing splits. Results: The proposed CNNAttLSTM model achieved 99% classification accuracy, outperforming the baseline CNN (86%) and CNN–LSTM (98%) models. It required only 3.7 million parameters, trained in 45 minutes on an NVIDIA Tesla T4 GPU, and achieved an inference time of 22 milliseconds per image, demonstrating high computational efficiency. The patch-based pseudo-temporal approach improved spatial–temporal feature representation, enabling the model to distinguish subtle differences between visually similar disease classes. Discussion: Results show that combining spatial feature extraction with temporal modeling and attention significantly enhances robustness and classification performance in plant disease detection. The lightweight design enables real-time and edge-device deployment, addressing a major limitation of existing deep-learning techniques. The findings highlight the potential of CNNAttLSTM for scalable, efficient, and accurate agricultural disease monitoring and broader precision agriculture applications.

CNNAttLSTM: an attention-enhanced CNN–LSTM architecture for high-precision jackfruit leaf disease classification.

Chronotype and Cancer: Emerging Relation Between Chrononutrition and Oncology from Human Studies.

Clinical phenotypes and short-term outcomes based on prehospital point-of-care testing and on-scene vital signs.

A Comparison of the Clinical Characteristics of Short-, Mid-, and Long-Term Mortality in Patients Attended by the Emergency Medical Services: An Observational Study.

Correction: Enhancing fault detection in new energy vehicles via novel ensemble approach.

Deep image features sensing with multilevel fusion for complex convolution neural networks & cross domain benchmarks.

Detecting hate in diversity: a survey of multilingual code-mixed image and video analysis.

DiabSense: early diagnosis of non-insulin-dependent diabetes mellitus using smartphone-based human activity recognition and diabetic retinopathy analysis with Graph Neural Network.

Mango is one of the most beloved fruits and plays an indispensable role in the agricultural economies of many tropical countries like Pakistan, India, and other Southeast Asian countries. Similar to other fruits, mango cultivation is also threatened by various diseases, including Anthracnose and Red Rust. Although farmers try to mitigate such situations on time, early and accurate detection of mango diseases remains challenging due to multiple factors, such as limited understanding of disease diversity, similarity in symptoms, and frequent misclassification. To avoid such instances, this study proposes a multimodal deep learning framework that leverages both leaf and fruit images to improve classification performance and generalization. Individual CNN-based pre-trained models, including ResNet-50, MobileNetV2, EfficientNet-B0, and ConvNeXt, were trained separately on curated datasets of mango leaf and fruit diseases. A novel Modality Attention Fusion (MAF) mechanism was introduced to dynamically weight and combine predictions from both modalities based on their discriminative strength, as some diseases are more prominent on leaves than on fruits, and vice versa. To address overfitting and improve generalization, a class-aware augmentation pipeline was integrated, which performs augmentation according to the specific characteristics of each class. The proposed attention-based fusion strategy significantly outperformed individual models and static fusion approaches, achieving a test accuracy of 99.08%, an F1 score of 99.03%, and a perfect ROC-AUC of 99.96% using EfficientNet-B0 as the base. To evaluate the model’s real-world applicability, an interactive web application was developed using the Django framework and evaluated through out-of-distribution (OOD) testing on diverse mango samples collected from public sources. These findings underline the importance of combining visual cues from multiple organs of plants and adapting model attention to contextual features for real-world agricultural diagnostics.

Dual-modality fusion for mango disease classification using dynamic attention based ensemble of leaf & fruit images.

Edge-Based Autonomous Fire and Smoke Detection Using MobileNetV2.

Efficient CNN architecture with image sensing and algorithmic channeling for dataset harmonization.

Enhanced interpretable thyroid disease diagnosis by leveraging synthetic oversampling and machine learning models.

Introduction: Weather classification plays a crucial role in applications such as environmental monitoring, disaster management, and smart city infrastructure. Accurate and efficient classification of weather conditions from images remains a challenging task due to variations in illumination, texture, and atmospheric conditions.Methods: This study proposes an efficient deep learning framework for multi-class weather classification by integrating the Xception architecture with Squeeze-and-Excitation (SE) blocks and a spatial attention mechanism. Transfer learning with pre-trained ImageNet weights was employed, and a comparative analysis was conducted using EfficientNet-B3, ResNet152V2, and Xception architectures. The proposed enhanced Xception model incorporates channel-wise recalibration and spatial feature refinement to improve representational capability. The model was trained and evaluated on the Multi-Class Weather Dataset (MWD), which consists of 1,125 images categorized into four classes: sunshine, cloudy, rain, and sunrise. To ensure robustness and generalization, 5-fold cross-validation, statistical significance testing, calibration analysis, and robustness evaluation under image perturbations were performed.Results: The proposed model achieved a classification accuracy of 99.06% on the test set. Additionally, it attained a macro precision of 98.3%, macro recall of 97.7%, and macro F1-score of 98.0%. The model demonstrated strong generalization capability and robustness under varying perturbation conditions, with only moderate computational overhead.Discussion: The integration of SE blocks and spatial attention significantly enhances feature representation by emphasizing informative channels and spatial regions. Compared to baseline architectures, the proposed framework shows superior performance in terms of accuracy and robustness. These results indicate that the model is well-suited for real-world weather classification applications, particularly in intelligent environmental monitoring systems.

Enhanced weather classification using xception with SENet and attention mechanisms.

Enhancing fault detection in new energy vehicles via novel ensemble approach.

Ensemble stacked model for enhanced identification of sentiments from IMDB reviews.

Evolving epidemiology, clinical features, and genotyping of dengue outbreaks in Bangladesh, 2000–2024: a systematic review.

Fundus image classification using feature concatenation for early diagnosis of retinal disease.

Human activity recognition (HAR) is essential in many applications, such as smart homes, assisted living, healthcare monitoring, rehabilitation, physiotherapy, and geriatric care. Conventional methods of HAR use wearable sensors, e.g., acceleration sensors and gyroscopes. However, they are limited by issues such as sensitivity to position, user inconvenience, and potential health risks with long-term use. Optical camera systems that are vision-based provide an alternative that is not intrusive; however, they are susceptible to variations in lighting, intrusions, and privacy issues. The paper uses an optical method of recognizing human domestic activities based on pose estimation and deep learning ensemble models. The skeletal keypoint features proposed in the current methodology are extracted from video data using PoseNet to generate a privacy-preserving representation that captures key motion dynamics without being sensitive to changes in appearance. A total of 30 subjects (15 male and 15 female) were sampled across 2734 activity samples, including nine daily domestic activities. There were six deep learning architectures, namely, the Transformer (Transformer), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Multilayer Perceptron (MLP), One-Dimensional Convolutional Neural Network (1D CNN), and a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) architecture. The results on the hold-out test set show that the CNN–LSTM architecture achieves an accuracy of 98.78% within our experimental setting. Leave-One-Subject-Out cross-validation further confirms robust generalization across unseen individuals, with CNN–LSTM achieving a mean accuracy of 97.21% ± 1.84% across 30 subjects. The results demonstrate that vision-based pose estimation with deep learning is a useful, precise, and non-intrusive approach to HAR in smart healthcare and home automation systems.

Human Activity Recognition in Domestic Settings Based on Optical Techniques and Ensemble Models.

A Hybrid Temporal-spectral Load Forecasting Model with Static Context Fusion for Smart Cities.

Infrared thermography to assess fatigue, injury risk factors and recovery in soccer: a systematic review of original studies.

An Integrated Machine Learning and Genomic Framework for Precise Detection of Gastric Cancer.

LwHM: lightweight hybrid classifier for SDN-attack detection using recursive feature elimination.

Novel Transfer Learning Approach for Detecting Infected and Healthy Maize Crop Using Leaf Images.

Organizational Culture Assessment Based on a Values-Based Coaching Program for Strategic Level Employees: The Case of GEDEME, Cuba.

Performance of the 4C and SEIMC scoring systems in predicting mortality from onset to current COVID-19 pandemic in emergency departments.

Prehospital acute life-threatening cardiovascular disease in elderly: an observational, prospective, multicentre, ambulance-based cohort study.

Pupilometer efficacy in monitoring anxiety in undergraduate medical students during high-fidelity clinical simulation.

Quantifying Domain-Specific Risk Signals in Lung Cancer Severity Prediction: A Multi-Domain Ablation Study Using XGBoost and SHAP.

Cardiovascular diseases are among the leading causes of mortality worldwide, with dietary factors being the main risk contributors. Diets rich in bioactive compounds, such as (poly)phenols, have been shown to potentially exert positive effects on vascular health. Among them, resveratrol has gained particular attention due to its potential antioxidant and anti-inflammatory action. Nevertheless, the results in humans are conflicting possibly due to interindividual different responses. The gut microbiota, a complex microbial community that inhabits the gastrointestinal tract, has been called out as potentially responsible for modulating the biological activities of phenolic metabolites in humans. The present review aims to summarize the main findings from clinical trials on the effects of resveratrol interventions on endothelial and vascular outcomes and review potential mechanisms interesting the role of gut microbiota on the metabolism of this molecule and its cardioprotective metabolites. The findings from randomized controlled trials show contrasting results on the effects of resveratrol supplementation and vascular biomarkers without dose-dependent effect. In particular, studies in which resveratrol was integrated using food sources, i.e., red wine, reported significant effects although the resveratrol content was, on average, much lower compared to tablet supplementation, while other studies with often extreme resveratrol supplementation resulted in null findings. The results from experimental studies suggest that resveratrol exerts cardioprotective effects through the modulation of various antioxidant, anti-inflammatory, and anti-hypertensive pathways, and microbiota composition. Recent studies on resveratrol-derived metabolites, such as piceatannol, have demonstrated its effects on biomarkers of vascular health. Moreover, resveratrol itself has been shown to improve the gut microbiota composition toward an anti-inflammatory profile. Considering the contrasting findings from clinical studies, future research exploring the bidirectional link between resveratrol metabolism and gut microbiota as well as the mediating effect of gut microbiota in resveratrol effect on cardiovascular health is warranted.

Resveratrol and vascular health: evidence from clinical studies and mechanisms of actions related to its metabolites produced by gut microbiota.

A Systematic Survey of AI Models in Financial Market Forecasting for Profitability Analysis.

Ultra Wideband radar-based gait analysis for gender classification using artificial intelligence.

Ultra-Wide Band Radar Empowered Driver Drowsiness Detection with Convolutional Spatial Feature Engineering and Artificial Intelligence.

Background/Objectives: The diet quality of younger individuals is decreasing globally, with alarming trends also in the Mediterranean region. The aim of this study was to assess diet quality and adequacy in relation to country-specific dietary recommendations for children and adolescents living in the Mediterranean area. Methods: A cross-sectional survey was conducted of 2011 parents of the target population participating in the DELICIOUS EU-PRIMA project. Dietary data and cross-references with food-based recommendations and the application of the youth healthy eating index (YHEI) was assessed through 24 h recalls and food frequency questionnaires. Results: Adherence to recommendations on plant-based foods was low (less than ∼20%), including fruit and vegetables adequacy in all countries, legume adequacy in all countries except for Italy, and cereal adequacy in all countries except for Portugal. For animal products and dietary fats, the adequacy in relation to the national food-based dietary recommendations was slightly better (∼40% on average) in most countries, although the Eastern countries reported worse rates. Higher scores on the YHEI predicted adequacy in relation to vegetables (except Egypt), fruit (except Lebanon), cereals (except Spain), and legumes (except Spain) in most countries. Younger children (p < 0.005) reporting having 8–10 h adequate sleep duration (p < 0.001), <2 h/day screen time (p < 0.001), and a medium/high physical activity level (p < 0.001) displayed a better diet quality. Moreover, older respondents (p < 0.001) with a medium/high educational level (p = 0.001) and living with a partner (p = 0.003) reported that their children had a better diet quality. Conclusions: Plant-based food groups, including fruit, vegetables, legumes, and even (whole-grain) cereals are underrepresented in the diets of Mediterranean children and adolescents. Moreover, the adequate consumption of other important dietary components, such as milk and dairy products, is rather disregarded, leading to substantially suboptimal diets and poor adequacy in relation to dietary guidelines.

Youth Healthy Eating Index (YHEI) and Diet Adequacy in Relation to Country-Specific National Dietary Recommendations in Children and Adolescents in Five Mediterranean Countries from the DELICIOUS Project.

A novel approach for disease and pests detection in potato production system based on deep learning.