Search Results (327)

This study presents a robust and extensible hybrid classification framework for accurately detecting diseases in citrus leaves by integrating transfer learning-based deep learning models with classical machine learning techniques. Features were extracted using advanced pretrained architectures—DenseNet201, ResNet50, MobileNetV2, and EfficientNet-B0—and refined via the minimum redundancy maximum relevance (mRMR) method to reduce redundancy while maximizing discriminative power. These features were classified using support vector machines (SVMs), ensemble bagged trees, k-nearest neighbors (kNNs), and neural networks under stratified 10-fold cross-validation. On the lemon dataset, the best configuration (DenseNet201 + SVM) achieved 94.1 ± 4.9% accuracy, 93.2 ± 5.7% F1 score, and a balanced accuracy of 93.4 ± 6.0%, demonstrating strong and stable performance. To assess external generalization, the same pipeline was applied to mango and pomegranate leaves, achieving 100.0 ± 0.0% and 98.7 ± 1.5% accuracy, respectively—confirming the model’s robustness across citrus and non-citrus domains. Beyond accuracy, lightweight models such as EfficientNet-B0 and MobileNetV2 provided significantly higher throughput and lower latency, underscoring their suitability for real-time agricultural applications. These findings highlight the importance of combining deep representations with efficient classical classifiers for precision agriculture, offering both high diagnostic accuracy and practical deployability in field conditions. Full article

Pine wilt disease, transmitted primarily by Monochamus alternatus (Hope, 1842) adults, causes severe ecological and economic losses globally. Conventional chemical controls face challenges of resistance and non-target toxicity. This study identified Beauveria bassiana (Bals.-Criv.) Vuill. strain B4 as a high-virulence biocontrol agent against adult M. alternatus. Laboratory bioassays compared four strains (B1–B4), with B4 exhibiting rapid lethality (LT₅₀ = 6.61 days at 1 × 10⁸ spores/mL) and low median lethal concentration (LC₅₀ = 9.63 × 10⁵ spores/mL). Critically, B4 infection induced significant behavioral suppression, including reduced appetite and mobility prior to death. In forest trials, pheromone-enhanced nonwoven fabric bags impregnated with B4 spores reduced trap catches by 66.4% within one month, with effects persisting for over a year without reapplication. The slow-release carrier system enabled continuous spore dissemination and sustained population suppression. These results demonstrate that B4’s dual action—rapid lethality and behavioral disruption—provides an effective, eco-friendly strategy for sustainable pine wilt disease management. Full article

The Global Trade Alert (GTA) website, managed by the United Nations, releases a large number of industrial policy (IP) announcements daily. Recently, leading nations including the United States and China have increasingly turned to IPs to protect and promote their domestic corporate interests. They use both offensive and defensive tools such as tariffs, trade barriers, investment restrictions, and financial support measures. To evaluate how these policy announcements may affect national interests, many countries have implemented logistic regression models to automatically classify them as either IP or non-IP. This study proposes ensemble models—widely recognized for their superior performance in binary classification—as a more effective alternative. The random forest model (a bagging technique) and boosting methods (gradient boosting, XGBoost, and LightGBM) are proposed, and their performance is compared with that of logistic regression. For evaluation, a dataset of 2000 randomly selected policy documents was compiled and labeled by domain experts. Following data preprocessing, hyperparameter optimization was performed using the Optuna library in Python 3.10. To enhance model robustness, cross-validation was applied, and performance was evaluated using key metrics such as accuracy, precision, and recall. The analytical results demonstrate that ensemble models consistently outperform logistic regression in both baseline (default hyperparameters) and optimized configurations. Compared to logistic regression, LightGBM and random forest showed baseline accuracy improvements of 3.5% and 3.8%, respectively, with hyperparameter optimization yielding additional performance gains of 2.4–3.3% across ensemble methods. In particular, the analysis based on alternative performance indicators confirmed that the LightGBM and random forest models yielded the most reliable predictions. Full article

Temperature and precipitation are the primary factors restricting litter decomposition in desert ecosystems. The desert ecosystems in Central Asia are ecologically fragile regions, and the climate shows a trend of “warm and wet” due to the regional climate change. However, the influencing mechanisms of warming and winter snow changes on litter decomposition are still poorly understood in desert ecosystems. Furthermore, the litter decomposition rate cannot be directly compared due to the large variations in litter quality across different ecosystems. Here, we simulated warming and altered winter snow changes in the field, continuously monitored litter decomposition rates of standard litter bags (i.e., red tea and green tea) and a dominant plant species (i.e., Erodium oxyrrhynchum) during a snow-cover and non-snow-cover period over five months. We found that warming and increased snow cover increased the litter decomposition rate of red tea, green tea, and Erodium oxyrhinchum, and had significant synergistic effects on litter decomposition. The effects of warming and winter snow changes on litter decomposition were more pronounced in April, when the hydrothermal conditions were the best. The decomposition rates of all three litter types belowground were higher than those on the soil surface, highlighting the important roles of soil microbes in accelerating litter decomposition. Furthermore, we found that warming and winter snow changes altered litter decomposition by influencing soil enzyme activities related to soil carbon cycling during the snow-cover period, while influencing soil enzyme activities related to soil phosphorus cycling during the non-snow-cover period. And, notably, decreased snow cover promoted soil enzyme activities during the snow-cover period. More interestingly, our results indicated that the decomposition rate (k) was the lowest, but the stability factor (S) was the highest in the Gurbantünggüt Desert based on the cross-ecosystem comparison using the “Tea Bag Index” method. Overall, our results highlighted the critical roles of warming and winter snow changes on litter decomposition. In future research, the consideration of relationships between litter decomposition and soil carbon sequestration will advance our understanding of soil carbon cycling under climate change in desert ecosystems. Full article

Storage of Blood units determines the accumulation of harmful substances, such as malondialdehyde (MDA) and free hemoglobin (fHb). These may lead to several complications, including cardiovascular, neurodegenerative, and metabolic disorders in recipients. The objective of this study was to evaluate the concentrations of MDA and fHb in canine leukoreduced (LR) and non-leukoreduced (NLR) packed red blood cells (pRBC) during the storage period of six weeks. Blood samples were collected from six healthy adult Weimaraner dogs (three females and three males). Whole blood was stored in citrate-phosphate-dextrose saline-adenine-glucose-mannitol additive solution (CPD-SAGM) bags and, for each donor, two pRBC units (one NLR and one LR) were produced and stored at 4 °C for 42 days. Samples were collected on days 0, 7, 14, 21, 28, 35, and 42, and analyzed for malondialdehyde (MDA) using a canine-specific ELISA method, and for free hemoglobin (fHb) using the Harboe direct spectrophotometric method. The results demonstrated a statistically significant reduction in MDA accumulation in LR-pRBC compared to NLR-pRBC blood units and lower values of fHb in LR at T6. However, no significant difference in fHb levels were demonstrated. These findings suggest that leukoreduction may limit oxidative stress during blood storage, reducing the potential adverse effects of transfusions related to oxidative damage. Full article

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that severely impacts motor function, necessitating early detection for effective management. However, current diagnostic methods are expensive and resource-intensive, limiting their accessibility. This study proposes a non-invasive, gait-based screening approach for PD using time-series analysis of video-derived motion data. Gait patterns indicative of PD are analyzed using videos containing walking sequences of PD subjects. The video data are processed via computer vision and human pose estimation techniques to extract key body points. Classification is performed using K-Nearest Neighbors (KNN) and Long Short-Term Memory (LSTM) networks in conjunction with time-series techniques, including Dynamic Time Warping (DTW), Bag of Patterns (BoP), and Symbolic Aggregate Approximation (SAX). KNN classifies based on similarity measures derived from these methods, while LSTM captures complex temporal dependencies. Additionally, Shapelet-based Classification is independently explored for its ability to serve as a self-contained classifier by extracting discriminative motion patterns. On a self-collected dataset (43 instances: 8 PD and 35 healthy), DTW-based classification achieved 88.89% accuracy for both KNN and LSTM. On an external dataset (294 instances: 150 healthy and 144 PD with varying severity), KNN and LSTM achieved 71.19% and 57.63% accuracy, respectively. The proposed approach enhances PD detection through a cost-effective, non-invasive methodology, supporting early diagnosis and disease monitoring. By integrating machine learning with clinical insights, this study demonstrates the potential of AI-driven solutions in advancing PD screening and management. Full article

Trailers for passenger cars are often used for the transportation of goods. There are various trailer designs. Most trailers are equipped with axles, which include swinging arms and are suspended by rubber segments. Observations have revealed that empty trailers have unfavorable driving properties when they are driven on uneven roads, for example, the wheels could jump off the road. Such a situation is dangerous because it is not possible to transmit any contact forces (longitudinal, lateral, or vertical) between the wheel and the road. The goal of the present research was to measure acceleration generated in a single-axle trailer when driving over a road obstacle. Measurements were conducted in a non-public area to avoid the risk of accidents. Acceleration was recorded using two accelerometers placed on the single-axle trailer frame above the wheels’ axle of rotation. Tests were performed using a vehicle–trailer combination at the chosen driving speeds, and the results for driving speeds of 20 and 30 km/h are presented. Wood plates with a height of 25 and 50 mm were used as an artificial road obstacle. The single-axle trailer was loaded with gravel bags weighing 0 to 300 kg. The measurements revealed that heavier trailer loads and lower driving speeds are safer for trailer operation. Furthermore, the measurements also demonstrated that the wheels were significantly more likely to jump off the road with a 0 kg load and low driving speed. Full article

Kiwifruit (Actinidia deliciosa) is a globally important crop presenting challenges for ensuring cross-pollination. This study aimed to (1) record the entomological fauna visiting flowers; (2) evaluate the visitation frequency of pollinators; and (3) test the use of lavender extract to enhance cross-pollination by honeybees and assess the impacts on fruit quality. Nine species of floral visitors were recorded as pollinators, although the most frequent were the exotic honeybee (Apis mellifera) and the native bees Bombus pauloensis and Xylocopa augusti. Honeybees increased their visitation to flowers when the attractant was used, improving pollination service and fruit quality compared to the control-bagged treatment, resulting in fruits that were 20 g heavier (115.4 g vs. 95.6 g, 95% CI). Similarly, the number of seeds per fruit and the fruit shape index (FSI) increased in treatments exposed to bee visitation when compared to the bagged control. However, differences in bee visitation among treatments suggested a non-linear relationship between bee activity and fruit quality. Nevertheless, achieving high-quality fruit standards across treatments could be explained by the extended floral lifespan, which allowed for a high number of visits and ensured pollination. Finally, we did not observe any bias in honeybee visitation by applying sugar syrup combined with the attractant. Hence, to increase honeybees’ visits to flowers, we recommend applying the scent directly in a water solution. Full article

Background: Mosquitoes from the Aedes (Ae.) genus are vectors of dengue, Zika, chikungunya, and other arboviruses, posing a significant public health threat. In 2005, Aedes aegypti was detected for the first time in Madeira Island, Portugal, in the city of Funchal, and has since become established in the region. In 2017, Aedes albopictus was detected for the first time in mainland Portugal. These invasion events require targeted entomological surveillance, which demands substantial human resources and a high management capacity for traditional vector monitoring. Following promising results obtained in laboratory conditions, a field-deployable model of a bioacoustic sensor for the automatic classification of mosquitoes integrated with a Biogents Sentinel trap as part of the VECTRACK system was tested in three regions in Portugal. Methods: The VECTRACK system was deployed in three locations: Funchal on Madeira Island, and Palmela and Algarve on mainland Portugal. Catch bags were manually inspected at intervals ranging from daily to weekly, resulting in a total of 38 captures in Madeira, 10 in Palmela, and 7 in the Algarve. Manual identifications were compared with those generated by the VECTRACK system, and the degree of correlation between the two datasets was assessed using Spearman’s rank correlation coefficient. Results: A total of 176 mosquitoes were captured in Madeira, 732 in Palmela, and 143 in the Algarve. Both manual and sensor-based identifications demonstrated similar performance, with high correlation observed between the two methods. Spearman’s rank correlation coefficients indicated high agreement for both female and male mosquitoes across all sites: Madeira: females = 0.84, males = 0.92, Palmela: females = 0.99, males = 0.84, Algarve: females = 0.98, and males = 0.99, all with p-values < 0.001. Conclusions: The VECTRACK system demonstrated strong performance in accurately distinguishing mosquitoes from non-mosquitoes, differentiating between Aedes and Culex genera, and identifying the sex of individual specimens. These promising results provide a solid foundation for the development of automated early warning systems and enhance mosquito surveillance strategies, which are critical for timely responses to potential vector-borne disease outbreaks. Full article

Dock door congestion is an essential and persistent concern within the realm of outbound warehouse logistics. The inability to accommodate outbound vehicles at the loading docks, especially during peak hours, disrupts internal warehouse operations, leads to bottlenecks, and contributes to substantial additional costs and delays. This paper addresses the critical issue of dock door congestion by proposing an integrated forecast–optimization framework for its prediction and management. The framework uses advanced forecasting methods and optimization techniques to increase warehouse throughput, boost operational efficiency, and predict potential congestion events using historical and real-time data. It combines two proven methodologies, maximum entropy bootstrap (MEB) and ensemble learning via bagging, with scenario-based stochastic optimization. This hybrid approach significantly improves upon traditional models by capturing the complex, non-monotonic components and multi-seasonality inherent in warehouse throughput data. Through a detailed real-world case study, we demonstrate how the proposed approach can accurately predict the number of trucks that can be serviced within specific time windows. This information is crucial for making operational decisions, such as whether to expand the warehouse. The approach can be generalized beyond the specific case study and offers valuable insights for any logistics or supply chain operation requiring the integration of stochastic optimization with predictive modeling. Full article

Hericium coralloides is a highly valued gourmet and medicinal species with growing market demand across East Asia, though industrial production remains limited by cultivation challenges. This study investigated the molecular characteristics, biological traits, domestication potential, and cultivation protocols of Hericium coralloides strains collected from the Changbaishan Nature Reserve (Jiling, China). Optimal conditions for mycelial growth included mannose as the preferred carbon source, peptone as the nitrogen source, 30 °C incubation temperature, pH 5.5, and magnesium sulfate as the essential inorganic salt. The fruiting bodies had a protein content of 2.43% g/100 g (fresh sample meter). Total amino acids comprised 53.3% of the total amino acid profile, while essential amino acids accounted for 114.11% relative to non-essential amino acids, indicating high nutritional value. Under optimized domestication conditions—70% hardwood chips, 20% cottonseed hulls, 8% bran, 1% malic acid, and 1% gypsum—bags reached full colonization in 28 days, with a 15-day maturation phase and initial fruiting occurring after 12–14 days. The interval between flushes was 10–12 days. The average yield reached 318.65 ± 31.74 g per bag, with a biological conversion rate of 63.73%. These findings demonstrate that Hericium coralloides possesses significant potential for edible and commercial applications. This study provides a robust theoretical foundation and resource reference for its artificial cultivation, supporting its broader industrial and economic utilization. Full article

Background: Pain is a complex phenomenon characterized by unpleasant experiences with profound heterogeneity influenced by biological, psychological, and social factors. According to the National Health Interview Survey, 50.2 million U.S. adults (20.5%) experience pain on most days, with the annual cost of prescription medication for pain reaching approximately USD 17.8 billion. Theranostic pain nanomedicine therefore emerges as an attractive analgesic strategy with the potential for increased efficacy, reduced side-effects, and treatment personalization. Theranostic nanomedicine combines drug delivery and diagnostic features, allowing for real-time monitoring of analgesic efficacy in vivo using molecular imaging. However, clinical translation of these nanomedicines are challenging due to complex manufacturing methodologies, lack of standardized quality control, and potentially high costs. Quality by Design (QbD) can navigate these challenges and lead to the development of an optimal pain nanomedicine. Our lab previously reported a macrophage-targeted perfluorocarbon nanoemulsion (PFC NE) that demonstrated analgesic efficacy across multiple rodent pain models in both sexes. Here, we report PFC-free, biphasic nanoemulsions formulated with a biocompatible and non-immunogenic plant-based coconut oil loaded with a COX-2 inhibitor and a clinical-grade, indocyanine green (ICG) near-infrared fluorescent (NIRF) dye for parenteral theranostic analgesic nanomedicine. Methods: Critical process parameters and material attributes were identified through the FMECA (Failure, Modes, Effects, and Criticality Analysis) method and optimized using a 3 × 2 full-factorial design of experiments. We investigated the impact of the oil-to-surfactant ratio (w/w) with three different surfactant systems on the colloidal properties of NE. Small-scale (100 mL) batches were manufactured using sonication and microfluidization, and the final formulation was scaled up to 500 mL with microfluidization. The colloidal stability of NE was assessed using dynamic light scattering (DLS) and drug quantification was conducted through reverse-phase HPLC. An in vitro drug release study was conducted using the dialysis bag method, accompanied by HPLC quantification. The formulation was further evaluated for cell viability, cellular uptake, and COX-2 inhibition in the RAW 264.7 macrophage cell line. Results: Nanoemulsion droplet size increased with a higher oil-to-surfactant ratio (w/w) but was no significant impact by the type of surfactant system used. Thermal cycling and serum stability studies confirmed NE colloidal stability upon exposure to high and low temperatures and biological fluids. We also demonstrated the necessity of a solubilizer for long-term fluorescence stability of ICG. The nanoemulsion showed no cellular toxicity and effectively inhibited PGE2 in activated macrophages. Conclusions: To our knowledge, this is the first instance of a celecoxib-loaded theranostic platform developed using a plant-derived hydrocarbon oil, applying the QbD approach that demonstrated COX-2 inhibition. Full article

In Brazil, Plasmodium infections in non-human primates (NHPs) have been associated with P. simium and P. brasilianum, which are morphologically and genetically similar to the human-infecting species P. vivax and P. malariae, respectively. Surveillance and monitoring of wild NHPs are crucial for understanding the distribution of these parasites and assessing the risk of zoonotic transmission. This study aimed to detect the presence of Plasmodium spp. genetic material in Platyrrhini primates from 47 municipalities in the states of Bahia and Minas Gerais. The animals were captured using Tomahawk-type live traps baited with fruit or immobilized with tranquilizer darts. Free-ranging individuals were chemically restrained via inhalation anesthesia using VetBag^® or intramuscular anesthesia injection. Blood samples were collected from the femoral vein. A total of 298 blood and tissue samples were collected from 10 primate species across five genera: Alouatta caraya (25), Alouatta guariba clamitans (1), Callicebus melanochir (1), Callithrix geoffroyi (28), Callithrix jacchus (4), Callithrix kuhlii (31), Callithrix penicillata (175), Callithrix spp. hybrids (15), Leontopithecus chrysomelas (16), Sapajus robustus (1), and Sapajus xanthosthernos (1). Molecular diagnosis was performed using a nested PCR targeting the 18S small subunit ribosomal RNA (18S SSU rRNA) gene, followed by sequencing. Of the 298 samples analyzed, only one (0.3%) from Bahia tested positive for Plasmodium brasilianum/P. malariae. This represents the first detection of this parasite in a free-living C. geoffroyi in Brazil. These findings highlight the importance of continued surveillance of Plasmodium infections in NHPs to identify regions at risk for zoonotic transmission. Full article

Background/Objectives: The SARS-CoV-2 and influenza A (H1N1)pdm09 pandemics have resulted in high numbers of ICU admissions, with high mortality. Identifying risk factors for ICU mortality at the time of admission can help optimize clinical decision making. However, the risk factors identified may differ, depending on the type of analysis used. Our aim is to compare the risk factors and performance of a linear model (multivariable logistic regression, GLM) with a non-linear model (random forest, RF) in a large national cohort. Methods: A retrospective analysis was performed on a multicenter database including 8902 critically ill patients with influenza A (H1N1)pdm09 or COVID-19 admitted to 184 Spanish ICUs. Demographic, clinical, laboratory, and microbiological data from the first 24 h were used. Prediction models were built using GLM and RF. The performance of the GLM was evaluated by area under the ROC curve (AUC), precision, sensitivity, and specificity, while the RF by out-of-bag (OOB) error and accuracy. In addition, in the RF, the im-portance of the variables in terms of accuracy reduction (AR) and Gini index reduction (GI) was determined. Results: Overall mortality in the ICU was 25.8%. Model performance was similar, with AUC = 76% for GLM, and AUC = 75.6% for RF. GLM identified 17 independent risk factors, while RF identified 19 for AR and 23 for GI. Thirteen variables were found to be important in both models. Laboratory variables such as procalcitonin, white blood cells, lactate, or D-dimer levels were not significant in GLM but were significant in RF. On the contrary, acute kidney injury and the presence of Acinetobacter spp. were important variables in the GLM but not in the RF. Conclusions: Although the performance of linear and non-linear models was similar, different risk factors were determined, depending on the model used. This alerts clinicians to the limitations and usefulness of studies limited to a single type of model. Full article

Background/Objectives: Deep models have achieved remarkable progress in the diagnosis of skin lesions but face two significant drawbacks. First, they cannot effectively explain the basis of their predictions. Although attention visualization tools like Grad-CAM can create heatmaps using deep features, these features often have large receptive fields, resulting in poor spatial alignment with the input image. Second, the design of most deep models neglects interpretable traditional visual features inspired by clinical experience, such as color–texture and edge features. This study aims to propose a novel approach integrating deep learning with traditional visual features to handle these limitations. Methods: We introduce the edge- and color–texture-aware bag-of-local-features model (ECT-BoFM), which limits the receptive field of deep features to a small size and incorporates edge and color–texture information from traditional features. A non-rigid reconstruction strategy ensures that traditional features enhance rather than constrain the model’s performance. Results: Experiments on the ISIC 2018 and 2019 datasets demonstrated that ECT-BoFM yields precise heatmaps and achieves high diagnostic performance, outperforming state-of-the-art methods. Furthermore, training models using only a small number of the most predictive patches identified by ECT-BoFM achieved diagnostic performance comparable to that obtained using full images, demonstrating its efficiency in exploring key clues. Conclusions: ECT-BoFM successfully combines deep learning and traditional visual features, addressing the interpretability and diagnostic accuracy challenges of existing methods. ECT-BoFM provides an interpretable and accurate framework for skin lesion diagnosis, advancing the integration of AI in dermatological research and clinical applications. Full article