Search Results (2,062)

Facial expression recognition (FER) in naturalistic settings is constrained by label ambiguity and variability in stimulus–response alignment. Adopting a data-centric perspective, this study examined whether emotional intelligence (EI)-stratified training data influence FER performance by treating EI as a qualitative factor associated with affective data consistency. Naturally elicited facial expressions were collected in a controlled emotion induction experiment with subjective arousal and valence ratings. Using response-driven labeling, neutral ratings were retained as indicators of ambiguity. Participants were grouped into High and Low EI based on the alignment between subjective evaluations and outputs from a pretrained affect estimator. Identical binary classifiers for arousal and valence recognition were trained while varying only the training data composition and evaluated across baseline, unambiguous, and ambiguous test sets using independent training repetitions with repetition-level statistical aggregation. EI-stratified training was associated with statistically detectable, context-dependent performance differences: group effects were observed primarily under baseline conditions and, to a lesser extent, under ambiguous conditions, whereas no reliable differences emerged under unambiguous conditions. Pooled discrimination differences were modest, but item-level analyses identified significant differences in classification correctness in specific task–condition combinations. Comparable patterns were observed across alternative backbone architectures. These findings indicate that FER performance in naturalistic contexts is influenced not only by model architecture but also by the statistical structure and internal coherence of the training data, supporting EI-informed data selection in ambiguity-prone scenarios. Full article

Background: Pediatric oral mucosal lesions are common and may indicate local or systemic disease, yet their recognition in primary healthcare often depends on non-dental professionals. Aim: To assess the preparedness of dentists, pediatricians, and family/general practitioners for pediatric oral mucosal conditions based on self-assessed diagnostic confidence, clinical management, and referral behavior. Methods: An online cross-sectional survey was conducted among 632 primary healthcare professionals (dentists: n = 262; family/general practitioners: n = 278; pediatricians: n = 92). The questionnaire assessed clinical exposure, self-assessed knowledge, diagnostic confidence, management practices, and referral patterns. Data were analyzed using chi-square or Fisher’s exact test and the Kruskal–Wallis test (p < 0.05). Results: Dentists reported significantly higher self-assessed knowledge and diagnostic confidence than pediatricians and family/general practitioners (p < 0.001). Good self-assessed knowledge of pediatric oral health was reported by 26.3% of dentists, compared with 7.9% of family/general practitioners and 6.5% of pediatricians. While most pediatricians (80.4%) and family/general practitioners (77.0%) reported routinely examining the oral cavity in children, independent treatment of oral mucosal lesions was more frequently reported by dentists (75.2%) than by pediatricians (52.2%) or family/general practitioners (70.9%) (p < 0.001). Referral patterns differed between groups, and willingness to attend future pediatric oral health education was high across all professionals (75.0–84.2%). Conclusions: Dentists demonstrated higher diagnostic confidence in pediatric oral mucosal lesions than pediatricians and family/general practitioners, who more often relied on referral. These findings support the value of targeted education and strengthened interdisciplinary collaboration in primary pediatric healthcare. Full article

Background: Alpine skiing imposes high biomechanical demands on the lower limbs, which are further amplified in individuals with transfemoral amputation due to prosthetic constraints. This study aimed to quantify three-dimensional knee flexion asymmetries during alpine skiing turns in transfemoral amputee skiers compared with non-disabled controls. Methods: Five unilateral transfemoral amputee skiers (intervention group) and five non-disabled ski instructors (control group) performed six left and six right turns on a skiing simulator under laboratory conditions. Knee flexion angles at the apex of each turn were analyzed using three-dimensional motion capture. Intra-individual differences between the prosthetic and intact limbs were assessed using paired comparisons, and inter-individual differences between groups were evaluated using independent statistical tests (p < 0.05), performed in IBM SPSS Statistics. Results: Intra-individual analysis revealed significant knee flexion asymmetries (p < 0.05) in almost all amputee participants at the apex of both left (mean difference = 7.74°, 95% CI: 3.38–12.09) and right turns (mean difference = 4.36°, 95% CI: 2.66–6.06). In the control group, asymmetries were smaller and reached significance only for the inside leg in both turns (mean difference = 4.02°, 95% CI: 2.51–5.54). Inter-individual comparisons demonstrated significant differences between the groups for both turning directions. During left turns (prosthetic limb on the inside), the largest difference was observed for the inside leg (26.9°, p < 0.001), while the smallest difference occurred for the outside leg (12.1°, p = 0.013). During right turns (prosthetic limb on the outside), the largest difference was found for the outside leg (19.0°, p < 0.001), with a smaller but still significant difference for the inside leg (14.0°, p < 0.001). Conclusions: Transfemoral amputee skiers exhibit a turning strategy that is qualitatively comparable to that of non-disabled skiers; however, it is characterized by a reduced knee flexion range of motion. These limitations appear to be primarily influenced by prosthesis mechanics and user-specific skill levels rather than by a fundamentally different movement strategy. Full article

Background/Objectives: Dental implants are an established modality for oral rehabilitation, but their biomechanical success depends on controlling peri-implant strain, which is influenced by implant angulation, splinting, and length. This in vitro study evaluated the effects of these variables on strain and displacement under axial and oblique loading using digital image correlation (DIC). Methods: Three CBCT-derived mandibular models were 3D-printed and restored with screw-retained full-metal crowns. Group 1 compared parallel vs. angulated implants; Group 2 assessed splinted vs. non-splinted restorations; and Group 3 compared short (4.2 × 6.25 mm) vs. long (4.2 × 13 mm) implants. All specimens were loaded to 500 N at 0°, 15°, and 30° using a universal testing machine. Strain and displacement were analyzed with Istra 4D software and statistically evaluated using ANOVA and independent t-tests (α = 0.05). Results: Parallel implants exhibited progressively higher strain with load angle, peaking at 30° (p < 0.01), while angulated implants recorded their highest strain at 0° (p = 0.008), indicating better adaptation to oblique forces. Splinted restorations significantly reduced strain at 0° and 30° (p = 0.023) and lowered displacement across all inclinations (p = 0.0001). Short implants consistently produced greater strain and displacement than long implants (p < 0.02). Conclusions: Angulated implants mitigated strain under off-axis loading compared to parallel configurations. Splinting decreased strain and displacement, while longer implants consistently improved biomechanical performance. Optimal selection of implant orientation, splinting, and length may minimize peri-implant strain under functional loads. Findings are limited to in vitro conditions with static loading and a single implant system. Full article

Automated electrocardiogram (ECG) arrhythmia classification remains challenging due to signal noise, inter-patient variability, and limited annotated data, which constrain the generalization of supervised learning approaches. This study presents a self-supervised ECG representation learning framework that combines contrastive pretraining with ensemble-based supervised classification. A signal-to-noise ratio criterion is applied during self-supervised pretraining to stabilize contrastive optimization, while all extracted ECG beats, including noisy segments, are retained during downstream evaluation. The learned representations are classified using a hybrid ensemble composed of convolutional encoders and tree-based models. Model evaluation follows strict patient-level partitioning with stratified 10-fold cross-validation and bootstrap-based uncertainty estimation on a held-out test set. Under this evaluation protocol, the framework achieved high beat-level performance on curated datasets (internal and external). Class-wise performance shows precision and recall values between 0.99 and 0.999 across normal, supraventricular, ventricular, fusion, and paced beat categories. External validation is conducted on independent ECG cohorts, including PTB-XL, Chapman–Shaoxing, and INCART 12-lead datasets. On these datasets, the hybrid model attains macro-F₁ scores ranging from 0.91 to 0.94, compared with standalone convolutional and handcrafted feature-based Random Forest classifiers evaluated under identical conditions. These results characterize the behavior of the proposed representation learning framework across heterogeneous patient populations and recording configurations. Full article

The complex coupling relationships among the thermal, mechanical, and electrical physical parameters of TiZrHf—based medium—entropy alloys represent a key factor restricting their practical applications under complex extreme environments. In this study, the thermo—mechanical—electrical coupling characteristics of TiZrHf and TiZrHfCu_0.8 medium—entropy alloys were systematically investigated using a self—developed experimental platform. The results demonstrate that TiZrHf and TiZrHfCu_0.8 alloys exhibit elastoplastic and superelastic—plastic compressive deformation behaviors, respectively, with both elastic modulus and ultimate strength decreasing monotonically with increasing temperature T. Electrical property measurements reveal that the electrical resistivities ρ of the two alloys range from 3 to 35 × 10⁻⁶ Ω·m. Notably, TiZrHfCu_0.8 possesses a lower resistivity that is independent of the test frequency f. Moreover, ρ increases with T but decreases with applied stress σ. At a frequency of 1 kHz, the real part of the relative dielectric constants ε_r of the alloys varies between −3.5 × 10⁸ and −0.5 × 10⁸ and increases with rising f, whereas the effects of T and σ on ε_r are opposite to those on ρ. Thermal property tests indicate that the thermal conductivities α of both alloys increase with T and eventually stabilize at 28.23 and 53.51 W·m⁻¹·K⁻¹, respectively, while the thermoelectric coefficients S are positively correlated with the heating rate, on the basis of comprehensive data analysis, multi—physical parameter (T, σ) dependent mathematical expressions for elastic modulus, strength, ρ, ε_r, α, and S were established, respectively. This work provides valuable insights into the material response mechanisms under complex service conditions, which are conducive to the optimization of alloy composition design and the promotion of their practical engineering applications. Full article

Soybean lodging severely impairs yield and quality, and its precise grading is a key prerequisite for intelligent agricultural management and loss assessment in agricultural insurance. Most existing studies have focused primarily on soybean lodging identification and lodging resistance evaluation, whereas methods for the precise differentiation of lodging grades remain to be refined. This study presents an improved AlexNet model integrated with a Local Feature Aggregation (LFA) attention mechanism and a dynamic optimization strategy for the accurate grading of soybean lodging. RGB imagery of soybean canopies during the grain-filling to early maturity stages was acquired via a multispectral unmanned aerial vehicle (UAV). A dynamic Dropout strategy was adopted to enhance model stability and mitigate overfitting, and the Particle Swarm Optimization (PSO) algorithm was employed to intelligently optimize key hyperparameters of the model. The results demonstrate that the optimized model achieved an overall accuracy of 94.23% on the test set, with an average loss of 0.0682 and an inference speed of 0.422 s/step. In independent field validation, the grading accuracies for the five lodging grades were 90.12%, 86.35%, 89.47%, 88.93%, and 92.76%, respectively, with a mean accuracy of 89.53%. The proposed model enables the rapid and precise grading of soybean lodging under field conditions, thereby providing effective technical support for intelligent field management and disaster loss assessment in soybean production. Full article

Automated analysis of liquid-based cervical cytology is increasingly supported by digital microscopy and deep learning. However, model generalization remains challenging due to scanner- and laboratory-induced domain shifts affecting color, texture, and morphology. In this study, we present a robust cell-level classification framework for liquid-based Pap smear cytology based on deep transfer learning, designed to operate under heterogeneous acquisition conditions. We construct a multi-source dataset by integrating three widely used public reference repositories (SIPaKMeD, Herlev, CRIC Cervix) with a proprietary cohort comprising 416 Whole Slide Images (WSIs) collected from two medical centers and digitized using different scanning systems. All labels are harmonized into four Bethesda categories (NILM, ASC-US, LSIL, HSIL), and cell-centered 224 × 224 patches are used as standardized inputs for model development and benchmarking. We evaluate state-of-the-art CNN backbones (ResNet50, EfficientNetB0, VGG16) and perform systematic ablation across data-source combinations to quantify robustness under acquisition variability. Among the evaluated models, ResNet50 yields the best overall performance on the independent test set (accuracy = 0.91; macro-F1 = 0.91), consistently outperforming EfficientNetB0 and VGG16. Importantly, incorporating proprietary multi-center WSI-derived data improves robustness to scanner-induced variation compared to training on public data alone. These findings demonstrate that combining diverse data sources can mitigate domain shift in cell-level cervical cytology classification. While clinically actionable screening requires slide-level aggregation (e.g., MIL-based WSI inference), the proposed classifier provides a robust component that can be integrated into end-to-end WSI screening pipelines in future work. Full article

Background/Objectives: Developmental prosopagnosia (DP) is a neurodevelopmental condition characterized by lifelong difficulties in face recognition. Although substantial work has examined identity-processing impairments in DP, less is known about whether these difficulties extend to other aspects of social cognition, including implicit trait judgements from faces. Prior research using Implicit Association Task (IAT) paradigms shows that neurotypical observers can automatically associate facial composites with personality traits such as extraversion. Although some studies report preserved explicit social evaluations in DP, to our knowledge, no previous work has assessed whether individuals with DP can form implicit personality trait impressions from faces. Methods: Using a cross-sectional experimental design, the present study examined whether adults with DP (N = 36) exhibit implicit extraversion trait associations, using a validated extraversion IAT online via Gorilla, following institutional ethics approval. Results: Group-level analyses showed a significant IAT effect, indicating sensitivity to congruent face–trait pairings. Single-case analyses using Crawford and Garthwaite’s modified t-test showed that no participant scored significantly below the normative neurotypical range. Conclusions: These findings indicate that implicit trait inference performance can remain within the normative range in DP despite severe identity recognition impairments, consistent with relative independence between social-evaluative and identity-related face-processing mechanisms. Full article

Membrane fouling remains a major obstacle to the performance of the reverse osmosis (RO) desalination processes. Artificial intelligence (AI) is now a promising approach for the reliable modeling of these complex systems. This study evaluates three modeling techniques—multiple linear regression (MLR), artificial neural networks (ANNs), and support vector regression (SVR)—for predicting transmembrane pressure (TMP) at the Boujdour desalination plant, based on five input parameters: temperature, turbidity, pH, conductivity, and feedflow. The analysis is based on an original dataset of 195 daily measurements, and due to the absence of timestamps, the study focuses on state-to-TMP prediction rather than chronological forecasting, with no temporal generalization claimed. Approximately 2000 augmented training samples generated using a conservative SMOGN approach were used for model development, while performance evaluation relied exclusively on 39 independent real test observations. Two modeling strategies were adopted: (i) a minimalist approach based on significant variables identified by an ordinary least squares (OLS) model (pH and conductivity), and (ii) a multivariate approach integrating all parameters to capture non-linear interactions. A rigorous validation framework was put in place to avoid information leakage and ensure the robustness and generalizability of the models. Performance was evaluated using R², RMSE, and MAE metrics, supplemented by robustness and significance analyses including bootstrap confidence intervals, paired statistical comparisons, and interpretability analyses based on permutation importance, partial dependence plots (PDPs), and individual conditional expectation (ICE) curves. The results indicate that the SVR model achieves the best average predictive accuracy among the tested models, albeit with moderate explanatory power. Full article

Background/Objectives: Tuberculosis and COVID-19 are two major infectious diseases with significant inflammatory and immunological impact on infected hosts and both conditions are independently associated with a prothrombotic state. However, evidence regarding their combined effect on in-hospital thrombotic risk remains limited. In this study, we aimed to explore whether patients with tuberculosis and COVID-19 coinfection are at a higher risk of developing thrombotic events during hospitalization than patients diagnosed with tuberculosis alone. Materials and Methods: We performed a retrospective, single-center cohort study, including adults hospitalized at the Pulmonology Clinic, Adult Tuberculosis ward of Mures County Clinical Hospital, between 2021 and 2023. Two groups were analyzed: patients with pulmonary tuberculosis who developed COVID-19 during hospitalization (n = 40) and patients with pulmonary tuberculosis without documented SARS-CoV-2 infection (n = 40). Demographic, clinical, laboratory, and imaging data were extracted from medical records. Padua and IMPROVE-DD scores were calculated retrospectively, a rapid mini-score was evaluated exploratorily. Comparisons between groups were performed using appropriate statistical tests and unadjusted odds ratios (ORs) with 95% confidence intervals (CIs) were reported. Given the limited number of events, an age-adjusted Firth penalized logistic regression model was used for multivariable analysis. Results: Thrombotic events occurred more frequently in the tuberculosis and COVID-19 co-infection group (22.5% vs. 10%), although statistical significance was not reached (p = 0.22; OR = 2.61). Patients with coinfection had significantly higher proportions of elevated Padua scores (55% vs. 20%, p = 0.002; OR = 4.88), while IMPROVE-DD showed values near the conventional threshold for statistical significance (37.5% vs. 17.5%, p = 0.07). D-dimer values did not reach statistical significance (p = 0.07) and platelet counts were significantly higher in patients with tuberculosis only (p = 0.001). Mortality did not differ significantly between groups (15% vs. 10%, p = 0.73). In age-adjusted multivariable analysis, tuberculosis and COVID-19 coinfection remained associated with higher odds of thrombotic events, with wide confidence intervals. Conclusions: Patients with concomitant tuberculosis and COVID-19 showed a higher thrombotic risk profile (Padua score) and numerically higher rates of in-hospital thrombotic events, without reaching statistical significance. Findings should be interpreted as exploratory and hypothesis-generating. Larger prospective studies with systematic imaging and multivariable adjustment are needed. Full article

The objective of this study was to evaluate the accuracy of detecting crop damage caused by wild boar in rapeseed fields using UAV (unmanned aerial vehicle)-derived RGB (red, green and blue) imagery and deep learning segmentation models. The experiments were conducted on rapeseed crops at full maturity shortly before harvest in central-western Poland in 2021. Four convolutional neural network architectures—U-Net (U-shaped network), U-Net++, DeepLabV3+ (deep learning + labelling), and PSPNet (Pyramid Scene Parsing Network)—were benchmarked using two input configurations: RGB imagery alone and RGB combined with the topographic position index (TPI) derived from a digital surface model (DSM). Model performance was assessed using overall accuracy, F1-score (harmonic mean of precision and recall), and Intersection over Union (IoU), with class-specific metrics reported to provide a realistic evaluation of damaged-area detection. For RGB-only data, overall accuracy ranged from 0.957 to 0.972, while damaged-class F1 and IoU reached 0.752 and 0.603, respectively, for the best-performing model (U-Net). When RGB data were supplemented with TPI, overall accuracy and damaged-class metrics changed only slightly, indicating limited benefit from the topographic feature under these field conditions. Non-damaged crop areas were consistently well-classified (F1 > 0.977, IoU > 0.955). These results confirm that UAV-based RGB imagery enables reliable late-season assessment of wildlife-induced crop damage, and that reporting class-specific metrics in spatially independent test sets is essential for realistic performance evaluation. Full article

Muscular fitness is a key component of health and athletic performance, and isometric strength is a widely used indicator. This study reports an agreement-based validation of the Isometric Strength Measurement Device (ISOMETRO) for upper-limb isometric tension measurements under controlled laboratory conditions. Twenty-one healthy young amateur rock climbers (11 men and 10 women) performed four upper-limb tensile tests (shoulder adduction at 90°, shoulder adduction at 60°, shoulder extension at 90°, and elbow extension at 90°). Agreement with an independent criterion device was evaluated using a force plate, while a series-connected load cell was used as an internal consistency check of the measurement chain. Linear mixed-effects models showed that ISOMETRO strongly predicted force plate values (β = 0.999, SE = 0.002, p < 0.001), with a marginal R² > 0.99. Bland–Altman analysis indicated negligible bias (−0.08 N) and narrow limits of agreement (−4.97 to 4.81 N), and concordance was excellent (CCC ≥ 0.996). The series-connected load cell comparison also showed near-unity agreement (β = 0.998, SE = 0.003, p < 0.001), supporting internal measurement chain integrity. These findings support excellent agreement between ISOMETRO and force plate measurements for upper-limb tensile isometric testing along the vertical axis in young amateur rock climbers under controlled laboratory conditions. However, given the specific sample characteristics and the strictly vertical laboratory configuration, these results should not be generalized to other populations, joint angles, force directions, or non-laboratory environments without further validation. Further studies are needed to confirm performance in more diverse contexts and to establish reliability for repeated-measurement applications. Full article

Background/Objectives: Autism is a complex neurodevelopmental condition characterized by social and cognitive impairments, with growing evidence implicating neuroinflammation, disrupted autophagy, apoptosis, GABAergic dysfunction, and gut permeability in its pathophysiology. Thus, this study aimed to evaluate the independent and combined effects of intermittent fasting (IF) and the next-generation probiotic Akkermansia muciniphila on behavioral outcomes and molecular markers in prenatal valproic acid (VPA)-induced autism model. Methods: Male rat offspring were allocated into five groups (n = 8 per group): control, VPA, IF, probiotic, and IF + probiotic. The groups other than the control group were exposed to 500 mg/kg VPA prenatally to establish an autism model. Intermittent fasting (16:8 time-restricted feeding) and Akkermansia muciniphila (1 × 10⁹ cfu/day) were applied for 30 days. Behavioral tests (stereotypy, social interaction, memory, and anhedonia) were performed during the last eight days of the treatment period, and the rats were sacrificed the following day for collection of brain tissue and serum samples. Proinflammatory, apoptotic, autophagic, and GABAergic markers were measured in the prefrontal cortex and hippocampus, while zonulin levels were measured in the serum. Data were analyzed using one-way ANOVA followed by Tukey’s post-hoc test. Results: Prenatal VPA exposure worsened all behavioral and molecular parameters. All treatments improved stereotypy, social interaction, and memory, whereas anhedonia improved only in the combined treatment group. The treatments also decreased neuroinflammation and apoptosis-related imbalance while enhancing autophagy and GABAergic markers. In terms of apoptosis- and autophagy-related markers, the IF-only and probiotic-only treatments were effective in the prefrontal cortex, while the IF + probiotic treatment showed its effect in both brain regions. Lastly, all treatments were successful in alleviating elevated serum zonulin levels. Conclusions: Intermittent fasting and Akkermansia muciniphila alleviate VPA-induced behavioral and neurobiological impairments. The combined treatment, in particular, offers stronger and multi-targeted therapeutic potential. Full article

Performance differences between female youth handball players selected for national teams and non-selected peers are often linked to strength, speed, and power. This study aimed to compare the conditioning capacities and body composition of national and non-national youth handball players. The sample included 36 female players (17.13 ± 1.75 years), 18 national and 18 position-matched non-national players. Anthropometry, sprint and change in direction ability, vertical jump, upper- and lower-body strength, aerobic capacity, and body composition were assessed using standard tests and bioimpedance analysis. For normally distributed variables, an independent-samples t-test was applied, while for variables that did not meet the normality assumptions, the Mann–Whitney U test was used. Cohen’s d was used to assess effect size. National team players showed significantly greater jump height (p < 0.001, d = 1.408), higher relative peak power (p < 0.001, d = 1.644), and faster 20 m sprint times (p = 0.004, d = −1.037). No significant differences were found in body composition or the other measured parameters, although a moderate Yo-Yo IRL1 effect size suggests a potential practical advantage in aerobic capacity for national team players. These results indicate that explosive power and linear speed are key discriminators for youth national-team selection. Full article