Search Results (506)

This study evaluated the performance of the Mindray BC-60R differential WBC count (AD) in comparison with a manual method (MD) in rabbits. Diagnostic blood samples were analyzed within 1 h of collection, and MD was performed by counting 100 leukocytes. Method comparison was first performed in the complete dataset (n = 27) and subsequently in a morphologically restricted dataset (n = 23), after separating four samples with major leukocyte morphological abnormalities associated with abnormal scattergram patterns. In the morphologically restricted dataset, Spearman’s rank correlation between AD and MD was excellent for heterophils (r = 0.94, p < 0.05) and lymphocytes (r = 0.92, p < 0.05) and moderate for monocytes (r = 0.71, p < 0.05) and basophils (r = 0.71, p < 0.05), but not significant for eosinophils, due to insufficient spread of values. Passing–Bablok regression revealed no statistically significant constant or proportional bias for heterophils, lymphocytes, or monocytes, whereas basophils showed both. The Bland–Altman analysis showed a significant underestimation of heterophils and overestimation of lymphocytes and basophils, while monocyte and eosinophil biases were not significant. Abnormal scattergram patterns were indicative of morphologically abnormal leukocytes. The Mindray BC-60R analyzer showed good agreement with the manual differential for the major leukocyte populations, particularly heterophils and lymphocytes, in samples without major morphological abnormalities. Evaluation of blood smear in rabbits is required to identify the atypical cell population on a scatterplot. Full article

Newborn screening (NBS) for inborn errors of metabolism is challenged by high false-positive rates, which may lead to parental anxiety and increased healthcare costs associated with diagnostic follow-up. False-positive results often arise from changes in metabolite concentrations that mimic metabolic disorders as a consequence of differences in perinatal factors or nutritional status. To address this, Collaborative Laboratory Integrative Reports (CLIR) and the associated post-analytical tools (PATs) using multivariate interpretation and covariate-adjusted reference intervals may be used to improve specificity of NBS algorithms. In the current study, we examined whether CLIR can be applied to optimize the Dutch NBS program by reducing the false-positive rates. We developed and validated a CLIR-based PAT for CPT1 deficiency, GA-I, IVA and MSUD within the Dutch NBS program. Single-condition tools (SCTs) and multivariate approaches, including marker ratios and covariate adjustments, were evaluated for their ability to discriminate true- and false-positive referrals. For CPT1 deficiency, age-adjusted SCT combined with birthweight, location correction, and the C18:1/methionine ratio substantially reduced false positives. For GA-I, C3DC-based ratios improved specificity while preserving true-positive detection, potentially reflecting postnatal renal immaturity in some cases. For IVA, the dual scatter plot fully separated true- and false-positive referrals, highlighting the limitations of single-marker screening. For MSUD, differences between false-positive and true-positive cases were more pronounced, yet a similar number of false positives were still referred; valine-related markers contributed to false positives, while leucine and the Xle/Phe ratio better identified true positives. CLIR-based post-analytical tools enhanced NBS specificity through covariate-aware, multivariate interpretation. This provides important input for decision makers in both the Dutch NBS, as well as the NBS community worldwide. Full article

Doppler Light Detection and Ranging (LIDAR) systems and a mesoscale meteorological model coupled with computational fluid dynamics (CFD) for the monitoring of low-level wind and turbulence have been extensively applied for the Hong Kong International Airport. This study represents the first application in Hong Kong to apply such techniques for the exploration of providing meteorological support for the operation of Unmanned Aircraft Systems (UASs) in a sandbox project in Hong Kong. The flight route under consideration is between the western coast of Hong Kong Island and an outlying island called Lamma Island, with a sea channel in between. Based on the LIDAR observations in three different prevailing wind directions, low-level turbulence may arise from wind flow disruptions by natural terrain and human-made buildings. Simulations of the wind and turbulence are attempted using the GPU-based FastEddy, with the turbulent kinetic energy equation being used to output the eddy dissipation rate (EDR). Comparisons between observed and simulated fields showed broadly consistent patterns across wind speed, wind direction, and EDR. Quantitative validation yielded RMSE of 1.35 m/s for wind speed, 28.4° for wind direction, and 0.032 m²/s² for EDR, with corresponding R² values of 0.72, 0.48, and 0.07, respectively. However, point-to-point comparison as in the scatter plot of the two datasets is still challenging, due to low correlation for EDR. Nonetheless, FastEddy is found to shed preliminary insights to generate reasonable simulations of low-level winds and turbulence to support the operation of UASs for the cases under study. These findings should be considered preliminary and exploratory given the limited number of case studies analyzed. More cases would need to be studied to find out the performance of FastEddy in other meteorological conditions. Full article

Objectives: We sought to evaluate the early efficacy of faricimab (Vabysmo^®) in treating diabetic macular edema (DME) and to explore the predictive value of multi-slice optical coherence tomography (OCT) biomarkers for anatomical and visual outcomes. Methods: In this retrospective cohort study, 26 anti-VEGF-naive DME patients (36 eyes) treated with intravitreal faricimab were analyzed from baseline through 6 months of follow-up. Best-corrected visual acuity (BCVA) was recorded, while central subfield thickness (CST) and various OCT biomarkers were evaluated using multi-slice OCT quantitative analysis. Logistic and linear regression models were utilized to examine predictive factors, and scatter plots were employed to assess the correlation between anatomical improvement and functional visual gain. Results: Changes in CST and BCVA, along with the evolution and predictive power of OCT biomarkers—including vitreomacular interface (VMI), epiretinal membrane (ERM), disorganization of the retinal inner layers (DRIL), intraretinal cysts (IRCs), hyperreflective foci (HRF), hard exudates (HEs), large outer-nuclear-layer cavities (LONLCs), ellipsoid zone disruption (EZD), and subretinal fluid (SRF)—were assessed. Post-treatment CST demonstrated rapid and significant reduction, decreasing from 377.7 μm (95% CI: 347.0–408.4) at baseline to 312.8 μm (95% CI: 287.5–338.2; p < 0.0001) at month 3 and 291.5 μm (95% CI: 275.6–307.4; p < 0.0001) at month 6, with an approximately 48 μm reduction post-first injection. Overall intraocular pressure (IOP) and BCVA showed no statistically significant improvement. Baseline analysis indicated that EZD was significantly associated with older age (p = 0.029), worse initial BCVA (p = 0.005), and thicker CST (p = 0.002). After adjusting for initial CST in the linear regression model, we identified the presence of baseline HEs as the sole independent predictor of substantial anatomical improvement (B = 45.9, p = 0.002). Conclusions: Faricimab demonstrated rapid and significant early anatomical improvements. Baseline HEs independently predicted the extent of CST reduction, potentially reflecting the greater fluid burden of eyes with more severe barrier breakdown. Baseline EZD was observed exclusively among eyes that did not achieve complete anatomical remission, although the small number of EZD-positive eyes (n = 7) precludes firm conclusions. Full article

Reliable electric load forecasting is an important engineering problem for power-system planning, grid stability, and mission-critical energy management. This paper presents a symmetry-aware simulation and modeling framework for medium-range electric load forecasting under noisy and uncertain operating conditions. The proposed approach combines a Multi-Model Partitioning Filter (MMPF) with Nonlinear Autoregressive Exogenous (NARX) submodels, while two adaptive optimization strategies, genetic algorithm-based resource allocation (GARA) and Particle Swarm Optimization (PSO), are used to optimize the contribution weights of the parallel predictors. The modeling process uses real commercial power-system data and evaluates the forecasting framework over April–September 2025. To simulate realistic engineering disturbances, correlated symmetric Gaussian noise is injected into the testing phase under moderate and heavy noise scenarios. The cyclic symmetry of temporal variables, such as hours and months, is preserved through unit-circle encoding, while the symmetry and asymmetry of residual error symmetric distributions are examined through scatter plot analysis. As for the context of forecasting residuals as diagnostic signals, it is important to transfer symmetry properties that can be used to evaluate the behavior of optimized predictors, along with the cyclic encoding of inputs. This means that by implementing residual-symmetry analysis, the conclusion that GARA and PSO produce concentrated, balanced, and biased errors under moderate noise and heavily correlated noise conditions can be achieved. Finally, our results show that both GARA and PSO improve the robustness of the hybrid MMPF-NARX model, but PSO consistently achieves lower MAPE values, smoother convergence, and lower computational burden. The optimal configuration is obtained with nine NARX submodels, beyond which additional model complexity offers no meaningful performance gain. Overall, the study shows that symmetry-aware modeling, adaptive optimization, and noise-based simulation can support more reliable forecasting in modern power-system engineering applications. Full article

Rutile is an oxide mineral widely distributed in igneous, metamorphic, and sedimentary rocks; it crystallizes in the tetragonal system. Trace-element abundances in rutile are influenced by the host-rock composition, redox conditions and crystallization history. Rutile exhibits high refractive index, strong dispersion, and adamantine luster. Its enrichment in high field strength elements (HFSEs) can be used to trace its formation environment. Owing to its inclusion-poor, compositionally uniform characteristics, rutile is particularly well-suited to in situ U-Pb geochronology. By integrating spectroscopic analysis, trace-element geochemistry, and U-Pb geochronology, this study systematically characterizes nine rutile samples from Madagascar, Pakistan, and Brazil, establishing a multi-dimensional scheme for origin discrimination. Spectroscopic analyses reveal that the infrared reflection band near 670 cm⁻¹ varies systematically with provenance. It appears as a broad, strong band in Madagascar samples, becomes weaker and narrower in Brazilian samples, and is partially absent in Pakistani samples. The Eg Raman mode of Brazilian rutile is slightly left-shifted and exhibits lower intensity, indicating a distinct lattice strain state. Analyzed samples occupy well-separated compositional fields on Nb–V, V–Ta, Zr–Hf and Nb–Ta binary variation plots. Specifically, Pakistani samples are characterized by high Nb and Ta contents and relatively lower V contents than the Brazilian and Madagascar samples. Madagascar samples show pronounced W enrichment and very low Cr. Brazilian samples display elevated Cr, V, higher U contents and more radiogenic Pb isotope compositions. Zr–W systematics and Cr–Nb bivariate discrimination allow inference of geological genesis. The Madagascar rutile is of hydrothermal origin, whereas the Pakistani and Brazilian rutile are metamorphic, derived from felsic/pelitic and mafic protoliths, respectively. LA-ICP-MS U-Pb geochronology yields a lower-intercept age of 504 ± 13 Ma (MSWD = 1.1) for the Madagascar sample (MD-1). This concordant, low-common-Pb age corresponds to the Pan-African orogeny and suggests strong potential as an in-situ U-Pb dating reference material. The Brazilian and Pakistani samples yield lower-intercept ages of 486 ± 53 Ma and 36.8 ± 2.9 Ma, respectively. However, the larger data scatter precludes their use as reference materials. Full article

Groundwater is a critical resource for domestic, agricultural, and industrial use in the Gilgel Gibe and Dhidhessa catchments of southwestern Ethiopia, where volcanic aquifer systems are the main sources. However, groundwater quality in these catchments has been under pressure from anthropogenic activities such as population growth, land-use changes, and pollution driven by rapid development and poor resource management. This study investigates hydrogeochemical processes and evaluates groundwater quality in volcanic aquifers using hydrochemical analyses and a stable isotope approach applied to 115 water samples. The spatial distribution of various physicochemical and hydrogeochemical parameters shows a distinct contrast between the highland and lowland regions, indicating topography-driven variations in water quality and geochemical processes. In hand-dug wells, springs, and surface waters, the ionic order is Ca²⁺ > Na⁺ > Mg²⁺ > K⁺ and HCO₃⁻ > NO₃⁻ > Cl⁻ > SO₄²⁻, whereas deep wells show Na⁺ > Ca²⁺ > Mg²⁺ > K⁺ and HCO₃⁻ > Cl⁻ > SO₄²⁻ > NO₃⁻. The predominant groundwater type is Ca-HCO₃, followed by Na-HCO₃ and Ca-NO₃, with other types including Ca-Mg-HCO₃, Ca-Na-HCO₃, and Na-Ca-HCO₃. Water types of Ca-HCO₃ and Ca-Mg-HCO₃ dominate the upland areas, indicating relatively young groundwater with moderate total dissolved solids (TDSs) and enrichment in δ¹⁸O and δ²H, where highly mineralized Na-HCO₃ water types prevail in the deep aquifers of the lowland regions, where δ¹⁸O and δ²H are relatively depleted. Principal component analysis, cross-plots of major cations versus HCO₃⁻, and mineral stability diagrams indicate that aluminosilicate weathering and dissolution are the dominant processes controlling groundwater chemistry in the study area. The higher saturation index values observed in the deep wells indicate water closer to mineral equilibrium, suggesting more extended water–rock interaction relative to the shallow wells. The CO₂ partial pressures calculated using PHREEQC exceed atmospheric levels (~10^−3.5 atm), indicating sources from atmospheric influx, soil, or biogenic activity for most samples, and deeper sources such as mantle degassing may be found in a few deep wells. Scatter plots of Cl⁻ vs. SO₄²⁻ and Cl⁻ vs. NO₃⁻, associated with Ca(NO3)₂, NaNO₃, and CaCl₂ water types, suggest that anthropogenic inputs are the second major factor influencing the area’s water chemistry. Stable isotope analyses and hydrochemical data indicate that groundwater in the area primarily originates from local precipitation, with isotopic signatures reflecting strong groundwater–surface water interaction. These findings improve understanding of regional hydrogeochemistry and groundwater quality and help identify promising zones for sustainable groundwater development. This study provides valuable insights into groundwater resource management both in the study area and in regions sharing comparable geological contexts. Full article

This study investigates intraspecific morphometric variability in the archaeozoological assemblages from the multi-period site of Șoimuș-Teleghi (Hunedoara County, Romania), encompassing occupation phases from the Early Neolithic (Starčevo-Criș culture) to the Late Bronze Age. The new AMS radiocarbon dates obtained refine the chronological framework of the site. The results reveal pronounced morphometric variability in cattle (Bos taurus), with horncore flattening indices, dental dimensions, and postcranial measurements enabling the differentiation of males, females, and probable castrates—the latter suggesting traction-oriented management strategies. A notably large horncore from the Vinča level may indicate hybridisation between domestic cattle and aurochs (Bos primigenius). Sheep (Ovis aries) display an increase in estimated withers height from the Neolithic to the Late Bronze Age, potentially reflecting improved husbandry. The pig (Sus domesticus) scapula scatter plot indicates a female-dominated culling pattern, while postcranial elements suggest relatively robust forelimbs combined with more gracile distal limb segments. Among wild taxa, red deer (Cervus elaphus) and wild boar (Sus scrofa) exhibit clear sexual dimorphism and generally large body sizes, consistent with the selective hunting of prime adult males. Comparative analysis with contemporary assemblages from Transylvania, Banat, and Hungary situates the Șoimuș-Teleghi material within broader regional trends, while highlighting certain distinctive features, particularly the large dimensions of Early Neolithic cattle and the diachronic size increase observed in caprine. Full article

Quince (Cydonia oblonga Mill.) is an important fruit crop, yet a systematic quality evaluation framework is lacking. This study comprehensively characterized multiple germplasms from distinct production areas. An integrated strategy combining physicochemical analysis, FT-MIR (Fourier transform mid-infrared spectroscopy), electronic nose (E-nose), headspace solid-phase microextraction–gas chromatography–mass spectrometry (HS-SPME-GC-MS), and chemometrics was employed. Significant variations were observed among accessions: certain varieties were observed to exhibit the highest pectin (1.86%) and total phenolic content (143.40 mg/100 g), while others showed superior firmness and titratable acidity. β-Damascenone in one accession was found to have an exceptionally high odor activity value (OAV) of 265.05. Multivariate analysis, including PLS-DA and OPLS-DA, effectively discriminated among quince accessions, with PLS-DA achieving 100% classification accuracy, and tentatively identified 18 key markers (VIP > 1) for accession discrimination. Loading scatter plot analysis further validated the contribution of these markers to the separation between accessions. However, due to the limited sample size (n = 18) and partial confounding between cultivar and origin, these findings should be considered exploratory and require validation in larger independent studies. This work provides preliminary insights into the diversity and geographical patterns of quality and flavor traits in quince germplasm, offering a preliminary foundation for germplasm evaluation and targeted utilization. Full article

Alkali-activated recycled aggregate concrete (AARAC) offers a sustainable alternative to traditional concrete but suffers from complex, non-linear mechanical behavior that challenges conventional prediction methods. This study develops and compares five machine learning models, linear regression (LR), M5P, Random Forest (RF), K-Nearest Neighbors (KNN) and XGBoost, for predicting the compressive strength (Cs), flexural strength (Fs), splitting tensile strength (Ss), pull-out bond strength (PT), and water absorption (Wa%) of AARAC. A dataset of 360 experimental samples, incorporating natural aggregate, recycled concrete aggregate (RCA), cement block aggregate (CBA), water-to-cement ratio (W/C), alkaline treatment status, and slump, was used. Models were evaluated via train/test split (80/20) and 10-fold cross-validation using R², MAE, RMSE, and MAPE. Random Forest achieved the highest test R² (0.8736) and lowest test MAPE (1.418%) and XGBoost (R² = 0.8605, MAPE = 1.557%). KNN and M5P performed moderately, while LR was the weakest (R² = 0.6958, MAPE = 2.147%). All tree-based models exhibited overfitting, with training R² up to 0.98. Scatter plot analysis revealed systematic underprediction by RF for Cs (constant offset of ~2 MPa) and increasing bias for PT, Ss, and Wa% at higher values. XGBoost gave perfect predictions for PT and Wa% but underpredicted Cs and Fs. K-fold cross-validation confirmed XGBoost as the most robust (mean R² = 0.9844). Correlation analysis showed W/C strongly increases Wa% (r = 0.80) and decreases PT (r = −0.73); RCA negatively affects mechanical properties, while CBA and alkaline treatment improve them. The study concludes that ensemble tree models, particularly Random Forest, are superior for AARAC prediction, but systematic bias requires post hoc calibration. Full article

This study investigates the acoustic emission (AE) characteristics associated with the shear failure mode based on the principles of active waveguide monitoring for the bedding rock slopes. Physical simulation experiments were conducted to assess the AE response during the shear-induced failure process of active waveguide structures. The findings indicate that during the initial loading phase, the scatter points of the signals are concentrated within a relatively narrow range. As the shear stress exceeds 90% of the peak stress and approaches the failure stage, there is a significant increase in the AE count and a rise in the high-frequency signals. Additionally, the distribution range of signals in the parameter correlation plot expands progressively. With increasing shear stress, the AE count, amplitude, and energy also rise gradually. And the emergence of continuous high-frequency signals is noted. During the failure stage, numerous microcracks initiate and propagate within the specimen, with signal amplitudes ranging between 40 and 90 dB. The peak frequency range of the AE signals broadens, with high-frequency components mainly concentrated between 350 and 450 kHz. Loading tests conducted at shear displacement rates of 0.25–1.5 mm/min reveal a strong correlation between the AE count and the shear displacement rate. Furthermore, prior to the shear failure of the active waveguide structures, the AE count shows a positive correlation with shear displacement. After the shear failure of the waveguide structure specimens, the AE count gradually decreases from a higher level to a lower level, demonstrating a negative correlation with shear displacement. The active waveguide structure can monitor the internal deformation conditions of the bedding rock slope so as to provide some reference for the early warning research. In addition, quantitative statistical analysis and curve fitting are conducted on the relationship between AE statistical count and shear displacement under different loading rates. The measured data show good agreement with the fitted curves, and a distinct two-stage evolutionary pattern (positive correlation before peak and negative correlation after peak) is quantitatively identified. These results further enhance the reliability of using AE parameters for quantitative evaluation of shear failure characteristics and displacement rate effects in bedding rock slopes. Full article

Ground-penetrating radar (GPR) was utilized for subsurface soil investigation in the Yellow River Delta, aiming to provide a scientific basis for the remediation performance of saline soils. The study particularly focuses on the red clay layer, a typical and characteristic soil horizon in this region. GPR antennas with central frequencies of 400 MHz and 900 MHz were adopted to investigate shallow soils within 1 m of the ground surface across three experimental plots (pits, undisturbed soils, and tilled soils) and 18 scattered measurement sites, followed by systematic analysis and interpretation of the acquired GPR profiles. During data acquisition, reasonable survey lines were deployed across the patchy bare areas of cultivated lands covering the experimental plots and measurement points to collect raw GPR data. Meanwhile, subsurface soil data were collected via test pits and borehole sampling along the survey lines. Raw GPR data were further preprocessed and postprocessed to characterize soil horizons and interpret subsurface stratigraphic structures. Finally, the correlations between the relative dielectric permittivity, reflection coefficient, and reflected wave amplitude of each soil layer were systematically analyzed. The results demonstrate that the 400 MHz antenna enables effective identification of soil layers within 1 m depth, while the 900 MHz antenna provides high-resolution detection for soil layers above 0.5 m. The red clay layer presents a distinct strong-amplitude reflection on GPR profiles, and the average relative dielectric permittivity of soils across the study area reaches 30.57. GPR profiles reveal that soil horizons with an absolute reflection coefficient greater than 0.01 yield detectable continuous reflection signals and allow uninterrupted stratigraphic interpretation. An empirical formula was established to calculate soil relative dielectric permittivity from soil moisture content, with a correlation coefficient of 0.9173. However, this formula ignores the influences of soil salinity and other trace soil elements. This study realizes rapid and accurate characterization of the depth and thickness of shallow soil layers, providing technical support for soil remediation of saline–alkali land in the Yellow River Delta. The findings also provide a valuable reference for evaluating the remediation effects, optimizing arable land utilization, preventing and mitigating soil salinization risks, and promoting the sustainable economic development of the study area. Full article

Background/Objective: Ultrasound is the first-line imaging modality for thyroid nodule assessment; however, it remains highly operator-dependent and subject to interobserver variability. Artificial intelligence (AI)-based systems have been proposed to improve reproducibility, yet evidence regarding their agreement with clinician assessment—particularly at the level of individual sonographic features—remains limited. Importantly, most available studies evaluate concordance rather than true diagnostic accuracy against an independent reference standard. To evaluate agreement between an AI-based ultrasound system and expert clinician assessment in thyroid nodule evaluation, focusing on concordance of size measurements, agreement in sonographic feature classification, and exploratory diagnostic performance relative to cytological outcomes. Methods: This retrospective single-center study included 74 thyroid nodules from adult patients undergoing routine ultrasound examination. Archived ultrasound images were independently assessed by an experienced clinician and an AI-based system. Agreement for quantitative measurements was evaluated using Bland–Altman analysis, while categorical features were assessed using percent agreement and Cohen’s kappa coefficients. Calibration was examined using scatter plots with the line of identity. Cytological results, when available, were used as a non-uniform exploratory reference standard for diagnostic analyses. Exploratory diagnostic performance was assessed using receiver operating characteristic (ROC) curves and area under the curve (AUROC) estimates. Given the study design, analyses primarily reflect agreement and measurement concordance rather than true diagnostic accuracy. Results: AI-derived and clinician measurements demonstrated strong agreement across all dimensions, with minimal systematic bias and stable calibration patterns. A small but consistent underestimation of one measurement axis by approximately 1 mm was observed. For categorical features, agreement ranged from fair to moderate (κ = 0.196–0.368), with the highest concordance for echogenic foci and lowest for echogenicity. Exploratory analyses showed variable diagnostic discrimination, with the best performance observed for size measurements and selected sonographic features. Conclusions: AI-based ultrasound analysis demonstrates robust agreement with clinician assessment for quantitative thyroid nodule measurements, while agreement for categorical feature classification remains moderate and variable. The findings highlight that the present study evaluates concordance rather than definitive diagnostic accuracy, particularly given the lack of a uniform independent reference standard. These results support the role of AI as an assistive tool in thyroid ultrasound practice, improving measurement reproducibility while requiring ongoing clinician oversight for qualitative interpretation. Full article

Forecasting fine particulate matter with a diameter of 2.5 μm (PM_2.5) is critically important due to its adverse effects on human health and environmental sustainability. Although ground-based monitoring stations provide accurate measurements, their limited spatial coverage restricts large-scale PM_2.5 assessment, especially in complex urban regions. Consequently, aerosol optical depth (AOD) derived from satellite imagery, combined with advanced deep learning (DL) techniques, has emerged as an effective alternative by offering wide spatial coverage and rich spatio-temporal information. This paper proposed an optimization-driven fuzzy transformer–deep belief network (ODFT-DBN) for accurate PM_2.5 air pollution prediction. The proposed framework integrates a fuzzy inference module to model uncertainty and nonlinear environmental relationships, a transformer encoder to capture long-range spatio-temporal dependencies, and a DBN to extract hierarchical features and improve prediction robustness. In addition, a novel multi-objective gray wolf optimizer (NMOGWO) is employed to jointly optimize the model hyper-parameters and fuzzy membership functions. The proposed approach is implemented for the city of Tehran, Iran, using meteorological variables, topographical features, ground-based PM_2.5 measurements, and satellite-derived AOD data. The ODFT-DBN model is compared with several benchmark methods, including bidirectional encoder representations from transformers (BERT), transformer, long short-term memory (LSTM), gated recurrent unit (GRU), convolutional neural network (CNN), DBN, and extreme gradient boosting (XGBoost). Experimental results demonstrate that the proposed framework achieves superior predictive performance, attaining an R² value of 0.94 and root mean square error (RMSE) of 0.8 μg/m³. Scatter plot analyses indicate a strong agreement between predicted and observed PM_2.5 values, while the proposed model exhibits low variance, stable convergence behavior, and acceptable computational time. Overall, the results confirm the effectiveness, robustness, and practical applicability of the proposed ODFT-DBN framework for spatio-temporal PM_2.5 forecasting. Full article

Background: Currently, expected goal models are tools that enable quantitative analysis in the study of conventional sports, although they have seen very little application in the Paralympic context. Objective: To present a trained expected goals model for 5-a-side blind soccer games based on an analysis of 164 offensive plays by the national team that won first place at the 2022 IBSA Copa América. The novelty of this work lies in being, to our knowledge, the first expected goals (xG) model developed for Paralympic blind football (B1): conventional xG weights cannot be transferred directly because shooting in F5 is governed by auditory orientation, the absence of an offside rule, a smaller rebound-walled pitch, and fully blind executors, so a sport-specific, reproducible and SHAP-interpretable benchmark is required where none previously existed. Materials and Methods: The SHapley Additive exPlanations library was used to analyze the data via partial dependency plots, dependency scatter plots, waterfall plots, decision plots, and SHAP heatmaps. Additionally, ten machine learning algorithms were compared, including logistic regression, random forest, extra trees, gradient boosting, XGBoost, LightGBM, CatBoost, support vector machine, k-nearest neighbors, and multilayer perceptron, using a 70/30 stratification process with fivefold stratified cross-validation to define the main hyperparameters. Results: The most consistent model was CatBoost (F1 = 0.778; AUC-ROC = 0.913; AUC-PR = 0.828; MCC = 0.729; Brier = 0.072), which allowed for independent analysis and evaluation of the dataset. The five main offensive variables were determined to be (i) distance to the goal before the shot; (ii) lateral coordinate; (iii) absolute magnitude of the shooting angle; (iv) magnitude of the progression vector; (v) proximity to the side kickboard. However, none of these variables proved to be decisive in the tournament (n = 24), a characteristic that the model captured as a significant negative contribution from the opponent variable. Conclusions: The expected goals model considered for this study serves as a starting point for further analysis of tactical variables in 5-a-side soccer for the blind. Because the model was trained on a single team in a single tournament with few positive cases, these results should be read as preliminary, hypothesis-generating tactical insights rather than validated performance estimates, and require external validation before transfer to other teams or competitions. Full article