Browse Articles
Discover research articles across all indexed journals
Deep learning and red deer optimiser for automatic cardiovascular disease identification on magnetic resonance images
Abstract Magnetic Resonance Imaging (MRI) is a non-invasive imaging method that can give detailed visualization of the cardiac structures and blood flow, which is effective in diagnosis of cardiovascular diseases (CVDs). It has been proposed that the combination of deep learning (DL) with MRI has an improved ability to automatically identify cardiovascular anomalies by identifying intricate patterns in large-scale imaging data. In this study, an Automated Cardiovascular Disease Detection framework (ACVD-RDODL) is proposed, which combines deep learning with the Red Deer Optimiser (RDO). After image enhancement methods like Wiener Filtering (WF) and Dynamic Histogram Equalization (DHE), features are extracted using radiomics. An Attention-Based Convolutional Gated Recurrent Unit (ACGRU) network is considered to ensure proper classification and RDO is used to optimize the hyperparameters and improve the performance of the progress model. As experimental testing of a benchmark cardiac MRI dataset shows, the proposed method is greater to the existing approaches in terms of classification accuracy and computational efficiency.
Feature interaction graphs for exact interpretable learning solver selection: an empirical diagnostic study
Using lithium refining residue and highly reactive metakaolin to optimise cement replacement in ternary and quaternary blended systems: Taguchi-Grey relational analysis method
Abstract The declining availability of conventional supplementary cementitious materials (SCMs), particularly fly ash (FA) and ground granulated blast furnace slag (GGBFS), has created a need to identify alternative low-carbon binders for cement-based materials. In this context, emerging SCMs such as lithium refining residue, known as delithiated beta spodumene (DBS), and highly reactive metakaolin (HRM) offer potential for reducing cement consumption while maintaining engineering performance. This study investigates the optimisation of blended binder systems based on a reference binary blend of 75% ordinary Portland cement (OPC) and 25% FA, aimed at reducing embodied carbon (EC) while maintaining satisfactory flowability and compressive strength. Two series of experiments were conducted: the first incorporated FA, DBS, and HRM, and the second incorporated FA, GGBFS, and HRM. A Taguchi L9 orthogonal array was adopted to design the experimental programme and evaluate the effects of the selected parameters on flowability and compressive strength. Grey relational analysis was then used to determine the optimal combination of responses, while analysis of variance (ANOVA) was performed to assess the relative influence of the individual parameters. The results indicate that HRM had a strong influence on early-age strength development, while the DBS material investigated in this study showed promising performance as a partial replacement material within the selected mortar-scale blended binder system. The optimum mix within the selected experimental domain in the first series comprised 63.75% cement, 21.25% FA, 10% DBS, and 5% HRM, while the optimum mix in the second series comprised 60% cement, 20% FA, 15% GGBFS, and 5% HRM. The embodied carbon of the optimum binder mixes was 264 and 261 kgCO 2 e per 450 kg of binder, respectively, representing reductions of 35% and 36% compared with the pure GP cement system.
Pd/Ti3C2 nanohybrids as heterogeneous catalyst for efficient catalytic reduction of hazardous water pollutants at ambient conditions
Reducing catastrophic forgetting in CNNs for plant stress classification using continual learning
Relationship between cerebrospinal fluid neurobiomarkers and symptoms in older patients with cognitive impairment: an observational clinical study
Predicting children’s undernutrition and its association with women’s asset ownership in Ethiopia: spatial machine learning
Abstract Child undernutrition remains a critical public health concern, with prominent geographic and structural disparities in Ethiopia. Understanding how women’s asset share and spatial context interact to affect nutritional risk is vital for effective interventions. This study investigates the spatial patterns of child undernutrition, its association with women’s asset ownership and other covariates, and the predictive performance of spatial machine learning techniques in making predictions. Using the nationally representative ESPS 2021/22 for model development and internal validation through spatial cross-validation with a 70/30 split, alongside temporal validation employing ESPS I, 2018/19. The metrics were used to compare four machine learning algorithms (random forest, gradient-boosted trees, support vector machines, and geographical random forests) with spatial features to a spatial generalized linear mixed model. The semivariogram result showed a strong spatial dependence, which supported the use of spatial modeling. The women’s asset ownership index, children’s age, distance to nearest market, mothers’ age, distance to nearest road, household size, child sex, region, toilet type, mother’s religion, source of drinking water, and place of residence were significant factors. The results demonstrate a distinct inverse spatial correlation between women’s asset ownership and child undernutrition, suggesting that women’s economic empowerment functions as a protective structural part. The result of temporal validation showed a growing impact of geographic and contextual factors. Gradient-boosted trees and random forests with spatial features make it easy to model intricate nutritional outcomes in different settings. The results underline geographically targeted child nutrition programs and policies that bolster women’s economic autonomy and resilience. Integrating spatial analytics and Gradient-boosted trees and random forests machine learning techniques with gender‑sensitive socioeconomic factors can strengthen precise public health strategies to reduce child undernutrition.
Sensory acceptance of flavoured high-calcium tempeh (TempeCal) chips among women of reproductive age
Dominance, drivers and thresholds of DEN, ANA, and DNRA quantified using a global dataset of cross-ecosystems and machine-learning
Abstract Nitrogen retention and loss in the critical zone are regulated by denitrification (DEN), anaerobic ammonium oxidation (ANA) and dissimilatory nitrate reduction to ammonium (DNRA), yet their global partitioning remains uncertain. A global dataset from 72 studies reporting DEN, ANA and DNRA rates plus environmental variables and functional genes was analyzed using random forests, generalized additive models and mixed-effects gene–rate analyses. Across ecosystems, contributions followed DEN (66.3%) > DNRA (21.9%) > ANA (11.8%). Wetlands showed consistently elevated DEN rates (mean 11.35 nmol N g⁻¹ h⁻¹) and the strongest ANA activity (mean 1.39 nmol N g⁻¹ h⁻¹), supporting moisture-rich systems as hotspots for DEN and ANA whereas highlands showed the highest relative DNRA (37.8% contribution; 6.89 nmol N g⁻¹ h⁻¹). Seven principal components explained 63.4% of the variance, highlighting critical zone organic matter pools, anaerobic metabolic intensity, and critical zone moisture and nutrient availability as the main environmental controls. Random forest models indicated that NO₃⁻ is the primary predictor for DEN and ANA (together with moisture and temperature/carbon supply), while DNRA is driven more strongly by NH₄⁺ (together with precipitation and temperature). DEN and ANA initially related to moisture negatively then positively at inflection points of about 15%. All processes were associated with hzsB , but correlations were weaker for DNRA. Together, these findings fill a key gap by providing a global, cross-ecosystem synthesis that jointly quantifies the rates, drivers, thresholds, and functional gene–rate associations of DEN, ANA, and DNRA.
Interpretable bearing fault diagnosis based on wavelet scattering network, PCA dimensionality reduction and PLUKAN network
Abstract Aiming at the issues of poor interpretability, weak generalization under small-sample conditions, and high industrial deployment cost in conventional deep learning-based bearing fault diagnosis models, this paper presents an interpretable fault diagnosis framework by integrating Wavelet Scattering Network (WSN), Principal Component Analysis (PCA), and Piecewise Linear Unit-based Kolmogorov–Arnold Network (PLUKAN). The framework adopts a three-stage collaborative architecture: robust feature extraction, efficient dimensionality compression, and structured interpretable classification. Multi-scale time-frequency scattering features with translation invariance and noise robustness are extracted via WSN, followed by efficient dimensionality reduction of high-dimensional features through PCA, and fault classification with structured decision interpretation is implemented by PLUKAN. Comprehensive validations are carried out on three public datasets and a self-built 6001-VS variable-speed bearing dataset. The results show that the framework achieves over 99% average accuracy under an 80% training ratio, corresponding to an 8:1:1 split, and demonstrates strong resilience under extreme few-shot conditions, maintaining competitive accuracy between 87.7% and 100% across four datasets when only 10% of the samples, corresponding to a 1:4.5:4.5 split, are used for training. Compared with the traditional KAN, PLUKAN reduces both the training time and inference time to approximately one-half, satisfying the real-time computing requirements of industrial edge devices. Meanwhile, by virtue of the physical interpretability of WSN features and the explicit mapping structure of PLUKAN, the framework overcomes the “black-box” limitation of deep learning models, and clearly establishes the correspondence between feature variations and fault mechanisms. The proposed WSN-PCA-PLUKAN framework realizes the collaborative optimization of diagnostic accuracy, computational efficiency and interpretability, providing a new engineering-feasible solution for the predictive maintenance of bearings in industrial rotating machinery.
Uniform volumetric encapsulation of antimicrobial agents into mesoporous silica for their slow release: the case of copper oxide
Abstract In the present work, a modified Stöber approach was developed for direct encapsulation of copper(II) tetraammine complex solution within the mesopores of SiO 2 particles during their formation. It was shown that a developed synthetic approach allows to uniformly distribute loaded compounds, avoiding their concentration close to the silica particle surface, which is common for post-synthetic techniques of loading. In turn, uniform distribution of copper species led to their slow and sustained release into the surrounding media, both for as prepared and calcined nanocomposites. It was shown that upon 24 h exposure to the solution, the release of copper species proceeds in almost linear fashion without any signs of rate decay. Both as prepared ([Cu(NH 3 ) 4 ] 2+ @SiO 2 ) and calcined (Cu x O@SiO 2 ) nanocomposites demonstrated antibacterial activity against the Gram-negative bacteria Escherichia coli and Pseudomonas aeruginosa , as well as the Gram-positive bacterium Staphylococcus aureus . However, the Cu x O@SiO 2 material appears to be more promising for antimicrobial applications due to its low cytotoxicity to human keratinocytes in vitro .
Design and analysis of modified octagonal ring-shaped MIMO antenna with connected ground for 5G sub-6 GHz n79, Wi-Fi 5, and Wi-Fi 6E band applications
Abstract The increasing demand for compact, high-isolation, and wideband antennas in modern wireless systems has created the need for efficient multiple-input multiple-output (MIMO) antenna designs suitable for 5G sub-6 GHz, Wi-Fi 5, and Wi-Fi 6E applications. MIMO antennas improve channel capacity, spectral efficiency, link reliability, and multipath performance, but maintaining compactness, wide bandwidth, low mutual coupling, and SAR compliance remains a major design challenge. This work proposes and experimentally validates a compact four-port modified octagonal ring-shaped MIMO antenna with a connected-ground structure for the 5G n79, Wi-Fi 5, and Wi-Fi 6E frequency bands. The antenna is designed on a low-cost FR4 substrate with dimensions of 60 × 60 mm². Four modified octagonal ring-shaped radiating elements are arranged orthogonally to improve polarization and spatial diversity. A connected-ground structure with multiple strips and slits is introduced to enhance impedance matching and reduce mutual coupling. The antenna is analyzed through design evolution, surface-current distribution, parametric optimization, S-parameters, far-field radiation characteristics, diversity parameters, and SAR evaluation using a human-head phantom model. The fabricated prototype is measured and compared with the simulated results. The proposed MIMO antenna achieves a wide impedance bandwidth of approximately 3.3–7.7 GHz, covering the 5G sub-6 GHz n79, Wi-Fi 5, and Wi-Fi 6E bands. Mutual coupling remains below − 17.5 dB across the operating band. The antenna exhibits stable radiation characteristics, with peak realized gain varying from about 4.3 dBi to 5.58 dBi and radiation efficiency between 80 and 96%. The measured and simulated results show good agreement. The diversity performance is satisfactory, with the envelope correlation coefficient close to zero, diversity gain nearly 10 dB, total active reflection coefficient below − 10 dB, channel capacity loss below the acceptable limit, and balanced MEG values. SAR values are also within the Federal Communications Commission (FCC) and International Commission on Non-Ionizing Radiation Protection (ICNIRP) safety limits for both 1 g and 10 g tissue models. The proposed connected-ground four-port MIMO antenna provides compact size, wide impedance bandwidth, high isolation, stable radiation performance, good diversity characteristics, and SAR compliance. Therefore, it is a promising candidate for 5G sub-6 GHz n79, Wi-Fi 5, Wi-Fi 6E, and other high-data-rate wireless communication systems.
Structural feature-based machine learning benchmarking for protein interface prediction
Winter-associated downregulation of ovarian NR5A2 correlates with impaired follicle development in the striped hamster (Cricetulus barabensis)
An efficient Pd embedded 2D g-C3N4 photo catalyst intramolecular cyclization reaction for synthesis of quinolin-fused benzo[d] azeto[1,2-a] benzimidazole analogues
Comprehensive bioinformatics analysis identifies KYNU as a novel Helicobacter pylori -associated biomarker with prognostic and therapeutic potential in gastric cancer
Abstract Helicobacter pylori ( H. pylori ) infection is a major risk factor for gastric cancer (GC), yet the key genes mediating this carcinogenesis remain unclear. This study aimed to identify H. pylori -related genes in GC and elucidate their molecular functions. Transcriptomic data from TCGA and GEO databases were analyzed using weighted gene co-expression network analysis (WGCNA), LASSO regression, and multivariate Cox analysis to construct a prognostic model. The immune landscape was also assessed. Validation involved RT-qPCR and immunohistochemistry (IHC). An in vitro H. pylori co-culture system was used to assess time-dependent changes in gene and protein expression via RT-qPCR, IHC, and Western blotting. A four-gene risk model (PDCD1 , KYNU , CYTL1 , and FZD2) was identified, demonstrating strong predictive capacity for overall survival as an independent prognostic factor. High-risk patients exhibited reduced immune cell infiltration. CellMiner analysis identified potential therapeutic agents targeting these genes. Among them, KYNU showed significant upregulation in GC tissues. Notably, in the co-culture system, KYNU expression markedly increased at both mRNA and protein levels following H. pylori infection in a time-dependent manner. The H. pylori -associated risk model represents a novel independent prognostic indicator for GC. Particularly, KYNU emerged as a pivotal gene in H. pylori -mediated GC, offering insights into disease progression and serving as a promising therapeutic target.