Search Results (1,345)

The increasing demand for real-time, energy-efficient machine learning (ML) models in edge and embedded scenarios highlights the limitations of traditional CPU-based inference pipelines. This work presents a pipeline that integrates Automated Machine Learning (AutoML) with High-Level Synthesis (HLS) to enable efficient deployment of ML predictors on Field-Programmable Gate Arrays (FPGAs). Using Auto-Weka for automated algorithm selection and Bayesian hyperparameter optimization, followed by model reimplementation and parameter extraction using scikit-learn and synthesis through the AMD/Xilinx Vitis HLS toolchain, the proposed workflow combines data-driven model exploration with hardware-oriented implementation. The pipeline adopts a hybrid, human-in-the-loop approach, reflecting current practical constraints in bridging heterogeneous software and hardware environments. Experimental results show up to a 9% latency reduction compared to CPU-based inference, more than a 7× throughput improvement through parallel FPGA instantiation, and more than an order of magnitude improvement in energy efficiency when evaluated in terms of energy per inference. The proposed methodology provides a reproducible and extensible workflow for integrating AutoML-based model discovery with FPGA deployment, while highlighting both the benefits of hardware acceleration and the remaining challenges in AutoML–hardware integration. Full article

Background/Objectives: End-stage renal disease (ESRD) profoundly affects patients’ daily functioning and well-being, particularly among individuals receiving long-term hemodialysis. Evaluating quality of life (QoL) is therefore essential for understanding the broader impact of ESRD and its treatment. This study aimed to assess QoL and identify demographic and treatment-related factors associated with QoL among patients undergoing maintenance hemodialysis at a regional dialysis center in Al Ahsa, Eastern Saudi Arabia. Methods: A cross-sectional study was conducted among 79 adult patients receiving maintenance hemodialysis for at least three months at King Abdulaziz Hospital, Al Ahsa. Quality of life was assessed using the WHOQOL-BREF questionnaire, with domain scores transformed to a 0–100 scale. Statistical analyses included descriptive statistics, independent t-tests, one-way ANOVA, Spearman’s correlation analysis, and multinomial logistic regression. Results: The physical domain demonstrated the lowest QoL score (49.69 ± 18.77), followed by the social domain (48.02 ± 24.92), whereas the psychological (63.83 ± 19.15) and environmental (67.56 ± 19.27) domains showed comparatively higher scores. Female sex, lower educational attainment, and night-shift dialysis were associated with poorer physical QoL. Significant positive correlations were observed among all QoL domains (p < 0.01), indicating strong interrelationships between different aspects of quality of life. Conclusions: Patients undergoing maintenance hemodialysis experienced substantial impairment in the physical and social dimensions of quality of life, while psychological and environmental well-being remained relatively preserved. These findings support the routine assessment of QoL in dialysis care and highlight the need for targeted interventions, including optimized dialysis scheduling, structured physical activity programs, and psychosocial support, to improve overall patient well-being with ESRD. Full article

Temporary floodplain ponds (TFPs) are short-lived water bodies forming in microtopographic depressions after flood recession and represent an important but poorly quantified component of floodplain hydrology. This study investigated the spatial and temporal dynamics of TFPs and their relationship with vegetation patterns in the natural floodplain of the Biebrza River, Poland. High-resolution TerraSAR-X data and Sentinel-2 multispectral imagery were combined with field-based vegetation surveys and statistical modeling. Threshold-based SAR classification showed that TFPs occupied more than 32% of the floodplain surface shortly after spring flood recession and stored, on average, over 250 L m⁻² of surface water, but disappeared within one month. Random Forest classification demonstrated that combining SAR and multispectral data improved overall vegetation mapping accuracy from 64.5% to 81.7% (Kappa from 0.574 to 0.780). A global chi-square test revealed a strong association between vegetation patterns and TFP occurrence (χ² = 224.9, p < 0.001, Cramér’s V = 0.40). Multinomial logistic regression identified TFP depth as the strongest predictor of plant community distribution. Rorippo-Agrostietum, Caricetum gracilis and Glycerietum maximae increased with TFP depth, whereas Alopecuretum pratensis and Phalaridetum arundinaceae declined. These results show that TFPs act as a fine-scale hydrological filter structuring floodplain vegetation mosaics and that SAR–optical data fusion is effective for detecting these transient habitat patterns. Full article

Background: Inappropriate antibiotic use for upper respiratory tract infections (URTIs) is common. However, existing research typically examines single indicators or broad categories of misuse practices, without considering the probabilistic and co-occurring nature of varying behaviours. This study aims to identify and characterise distinct antibiotic behavioural profiles within a general population to inform personalised interventions. Methods: This is a broadly representative population-based study of adult Singapore residents between November 2020 and January 2021. Latent class analysis was first performed to identify distinct profiles, followed by multinomial logistic regression to determine individual characteristics associated with each profile, with interaction effects examined. Results: Amongst 2004 respondents, the majority were “antibiotic appropriates” (53.0%), followed by “antibiotic avoiders” (24.9%), and “antibiotic seekers” (22.2%). “Antibiotic seekers” expected antibiotics for cold/flu, hopped between doctors to source antibiotics, used leftover antibiotics, stopped antibiotic courses prematurely and perceived antibiotics as harmless and useful for treating cold/flu. Individuals with poor knowledge of antibiotic use (AOR 3.71, 95% CI 2.89–4.76, p < 0.001) and antimicrobial resistance (AOR 3.51, 95% CI 1.05–11.76, p = 0.042), low eHealth literacy (AOR 1.34, 95% CI 1.02–1.77, p = 0.038), and high trust in doctors (AOR 2.14, 95% CI 1.44–3.17, p < 0.001) were more likely to be “antibiotic seekers”. Older adults with lower education levels were particularly likely to be “antibiotic seekers”. Conclusions: Approximately one-in-five Singapore residents are antibiotic seekers. Targeted education during multiple clinic visits, given the high trust in doctors, can address antibiotic knowledge gaps and misperceptions, reducing antibiotic misuse. Full article

Objective: This study aimed to evaluate the knowledge and clinical awareness of local anesthetic systemic toxicity (LAST), preventive practices, knowledge of lipid emulsion therapy, and institutional availability among dentists actively practicing in Türkiye, and to examine the demographic and professional factors associated with knowledge level. Methods: This descriptive cross-sectional study was conducted using a 15-item online questionnaire developed by the researchers. The analyses included 369 dentists actively practicing in Türkiye. Data were analyzed using descriptive statistics, the Kruskal–Wallis test, Dunn–Bonferroni pairwise comparisons, Spearman rank correlation, and multinomial logistic regression analysis. Results: Among the participants, 45.8% reported having basic knowledge of LAST, whereas only 2.7% reported detailed knowledge, including the management steps. Although 40.1% stated that they calculated the local anesthetic dose according to the patient’s body weight, only 3.3% reported preparing an emergency response plan for LAST, and 2.2% indicated that they were prepared to use treatment options such as lipid emulsion. Regarding lipid emulsion therapy, 59.1% of participants had low knowledge and 24.4% had superficial knowledge; only 0.5% reported detailed knowledge of the administration steps and dosing protocol. In terms of institutional availability, 45.0% did not know whether lipid emulsion was available at their institution, 40.9% reported that it was unavailable, and 14.1% reported that it was available. Knowledge levels differed according to professional status; however, the effect size was small (H(2) = 13.129; p = 0.001; ε² = 0.030). No statistically significant association was found between years of professional experience and knowledge level (ρ = 0.020; p = 0.702). Conclusions: Although dentists’ self-reported awareness of LAST varied, detailed knowledge of the administration steps and dosing protocol for lipid emulsion therapy, as well as institutional preparedness, remained limited. The findings suggest that strengthening practice-oriented education on the prevention and management of LAST and reviewing lipid emulsion availability and emergency response protocols in clinical institutions may be beneficial. Full article

Wildfire behavior can roughly be described as wind-, fuel-, or topography-driven, but these labels usually rest on expert judgment and post-fire analysis rather than on simple, reusable rules. Here, we build a small, interpretable classifier that predicts the dominant fire behavior driver: fuel-driven (plume/convection dominated), wind-driven, or topography-driven, from basic environmental information. The classifier is built using ERA5 10 m wind and relative humidity, summary elevation metrics, and fuel descriptors for a 14-event dataset of coastal Croatian wildfires. We compare the performance of multinomial logistic regression, random forest, and decision tree-based classifiers, focusing on agreement between their coefficients, feature importances, and splits rather than on formal optimization. All three models converge on a simple common rule set. Relative humidity, mean elevation, and elevation range emerge as the main axes of variation, consistent with basic fire behavior physics and published fire type schemes. Despite the small dataset, the classifier formalizes expert intuition in a transparent way and offers a template for scaling larger datasets, where it could evolve into a quick diagnostic of the dominant spread driver for ongoing fires. Full article

The Chang 8 Member of the Jinghe Oilfield in the Ordos Basin is a low-porosity, ultra-low-permeability reservoir with many faults and fractures, complex structures, and strong heterogeneity. Conventional logging curves do not clearly distinguish among different structural units, making it difficult to identify the internal structures of fault-fracture reservoirs. Current methods mainly use logging curves and rock mechanical parameters. In these reservoirs, experiments are costly, numerical simulations take a long time, and identification is often inefficient. To improve identification accuracy and efficiency, this study developed a two-layer Stacking ensemble model for the Chang 8 Member. The dataset was derived from conventional well-log data from five wells in the Chang 8 Member and contained 816 labelled depth samples. Among them, 569 original samples from wells A1, A2, and A3 were used for model development, while 247 samples from wells JH55P10 and JH2301H were reserved for independent well-level validation. In the first layer, a support vector machine (SVM), XGBoost, and a random forest (RF) were used as the base learners. The hyperparameters of the base learners were optimized using grid search and K-fold cross-validation. In the second layer, multinomial logistic regression was used as the meta-learner to integrate the class-probability outputs of the base learners and generate the final predictions. Individual models showed limitations in distinguishing the three internal structural units of fault-fracture reservoirs. By integrating the complementary outputs of the base learners, the Stacking model achieved an overall accuracy of 0.89, exceeding the accuracies of the individual models on the internal hold-out test set. The results indicate that the proposed framework can improve the accuracy and class balance of multi-class identification on the present dataset and provide a practical approach for the detailed evaluation of internal structural units in low-porosity, low-permeability fault-fracture reservoirs. Full article

Government-funded teacher education in China links financial support, teacher preparation, employment expectations, and post-graduation service obligations. Understanding how pre-service preschool teachers perceive this policy-based pathway is important for explaining their professional identity development; therefore, this study examined how policy satisfaction and perceived teacher support were associated with the professional identity of government-funded pre-service preschool teachers in Chongqing, Southwest China. Based on paper-based questionnaire data from 620 participants, Latent Profile Analysis identified four profiles: Dissatisfied–Low Support, Moderately Satisfied–Moderate Support, Highly Dissatisfied–Low Support, and Highly Satisfied–High Support. Multinomial logistic regression showed that only-child status and age significantly predicted profile membership, and one-way ANOVA and multiple regression further indicated that professional identity differed significantly across profiles, with lower scores observed in the less satisfied and less supported profiles after controlling for demographic covariates. These findings suggest that strengthening policy communication and accessible teacher support may help promote professional identity development among government-funded pre-service preschool teachers. Full article

Background: Inflammation and lipid metabolism play critical roles in coronary artery calcium (CAC) progression. This study aimed to investigate the relationship between neutrophil-to-high-density lipoprotein-cholesterol ratio (NHR) and CAC score. Methods: This cross-sectional study included 2193 eligible participants from Sichuan Provincial People’s Hospital between November 2015 and July 2025. The correlation between NHR and CAC score was evaluated using multivariable logistic and multinomial logistic regression models. Restricted cubic splines (RCS) were employed to assess potential nonlinear relationships. Sensitivity analyses and subgroup analyses were performed to test the robustness of the findings. Results: Among 2193 eligible participants, 64.89% had detectable CAC. Higher NHR levels were significantly associated with increased CAC prevalence. After adjustment for multiple confounders, each 1-unit increase in NHR was associated with 9.0% higher odds of CAC (odds ratio (OR): 1.09, [95% confidence interval (CI) 1.03–1.16], p = 0.002). Compared with the lowest NHR quartile, the highest quartile was associated with modestly higher odds of CAC (OR: 1.43 [95% CI 1.05–1.95], p = 0.022). In multinomial analyses, NHR was modestly but significantly associated with CAC across mild, moderate, and severe calcification stages. The RCS analysis showed a linear relationship between NHR and CAC. Subgroup analyses and sensitivity analyses confirmed the robustness of the findings. Conclusions: Elevated NHR was associated with an increased likelihood of CAC. As a simple and readily available marker, NHR may reflect inflammatory and lipid metabolic status related to subclinical atherosclerosis, although its clinical utility requires further confirmation. Full article

This pilot study investigated user-preferred ventilated seat intensity levels under simulated hot vehicle cabin cooling conditions to support adaptive seat ventilation systems. Thirty-three participants were exposed to a transient cooling environment in which cabin temperature decreased from approximately 38 °C to 25 °C following air-conditioning activation. Participants selected preferred seat ventilation intensity levels (Low, Medium, or High) while demographic and environmental variables were evaluated. Results indicated that cabin temperature was the strongest predictor of ventilation intensity preference, followed by elapsed cooling time and relative humidity. Age demonstrated a statistically significant effect, whereas height, body weight, BMI, and sex were not statistically significant predictors. An exploratory multinomial logistic regression model demonstrated preliminary predictive feasibility with a cross-validated classification accuracy of 73.2%. The findings suggest that occupant-preferred ventilation intensity levels may be estimated using demographic and environmental variables under transient vehicle cooling conditions. Full article

Background: Physical resilience (PR) plays a critical role in the functional recovery of older adults following fragility fractures, yet individual heterogeneity remains underexplored. Objective: To identify latent profiles of PR among older adults following fragility fractures and to examine their biopsychosocial predictors. Methods: From October 2025 to March 2026, 224 older adults with fragility fractures were recruited using convenience sampling from a tertiary hospital in China. Data were collected using demographic questionnaires, the Physical Resilience Instrument for Older Adults, Perceived Social Support Scale, and General Self-Efficacy Scale. Latent profile analysis identified profiles, followed by multinomial logistic regression examining biopsychosocial predictors. Results: Three distinct resilience profiles emerged: low resilience—physically and mentally vulnerable (31.7%); moderate resilience—limited adaptation (51.5%); and high resilience—potential activated (16.8%). Significant profile predictors included handgrip strength, nutritional risk, marital status, general self-efficacy, and perceived social support (p < 0.05). Conclusions: The distinct heterogeneity of PR among older adults with fragility fractures underscores the necessity for tailored, risk-stratified nursing. In clinical practice, interventions for the highly vulnerable low-resilience group should prioritize multidisciplinary nutritional optimization and early physical rehabilitation. For patients with moderate resilience, integrating spousal support and cognitive-behavioral strategies is crucial to enhance self-efficacy and prevent functional decline. For the high-resilience cohort, leveraging robust social support networks and empowerment-based strategies can maximize their intrinsic recovery potential. Full article

Background/Objectives: Sarcopenia significantly increases the risk of hospitalization and mortality in older adults. However, the prevalence by sex varies across populations and according to the criteria used by the European Working Group on Sarcopenia in Older People 2 (EWGSOP2). Hence, in the current study, we aim to characterize the prevalence of sarcopenia by jointly evaluating all three diagnostic domains in a Mexican cohort of older adults subscribed to a health insurance plan. Methods: We performed binomial and multinomial regression models, ordinal logistic regression, and multiple linear regression from the data corresponding to muscle mass (ASMI), muscle strength (handgrip dynamometry), and physical performance (SPPB) from 556 Mexican older adults (62.2% female; 72.27 ± 6.35 y.o.) categorized by sarcopenia severity according to the EWGSOP2. Results: Men exhibited significantly higher absolute muscle mass and strength across all categories (p < 0.001). Additionally, our statistical analyses demonstrate that age, but not sex, is involved in sarcopenia severity in this population. Moreover, when analyzing the disaggregated EWGSOP2 domains, the results demonstrate that sex is significantly associated with muscle mass (ASMI) and muscle strength (Handgrip strength). Conclusions: These results suggest that sex only influences both muscle mass (ASMI) and muscle strength (handgrip strength) while sarcopenia severity only depends on age. Full article

Background: Tuberculosis-related pulmonary changes may overlap radiologically and clinically with lung cancer, complicating diagnostic interpretation, staging, and therapeutic planning. This study evaluated the relationship between tuberculosis status, thoracic imaging patterns, and clinical characteristics in patients diagnosed with lung cancer. Methods: A retrospective cohort study was conducted between February 2020 and December 2025 at the Clinical Pneumophtisiology Hospital, Constanța, Romania. A total of 620 patients with lung cancer were analyzed. Patients were classified into three groups: no tuberculosis, post-tuberculosis sequelae, and active tuberculosis. Demographic, clinical, laboratory, histopathological, functional, and radiological variables were assessed. Associations between tuberculosis status and imaging findings were evaluated using chi-square testing, effect-size analysis, and multinomial logistic regression. Results: Post-tuberculosis sequelae were identified in 337 patients (54.4%), active tuberculosis in 51 patients (8.2%), and no tuberculosis-related disease in 232 patients (37.4%). Adenocarcinoma was the most frequent histological type, occurring in 359 patients (57.9%). Significant associations with tuberculosis status were observed for fibrotic/interstitial or bronchial changes, infectious-inflammatory changes, cavitary/destructive lesions, atelectatic/retractile changes, pulmonary opacities, pleural involvement, and mediastinal/hilar adenopathy. The strongest effects were found for fibrotic/interstitial changes, infectious-inflammatory changes, and cavitary/destructive lesions. In regression analysis, active tuberculosis was most strongly associated with infectious-inflammatory changes, cavitary lesions, pulmonary opacities, and fibrotic abnormalities, while post-tuberculosis sequelae were mainly associated with chronic fibrotic and structural pulmonary changes. Conclusions: Tuberculosis-related abnormalities frequently coexist with lung cancer and may mimic or obscure malignant findings. Recognition of these overlapping patterns is essential for accurate radiological interpretation and individualized clinical management. Full article

Background/Objectives: Healthcare workers (HCWs) were the most exposed job category to SARS-CoV-2, due to patient care, HCW-to-HCW transmission, and community exposure. However, the relative relevance of each source is still debated. To address this issue, this study investigated the dynamics of the professional and extra-professional determinants of infection across the pandemic among a large, multicenter cohort of HCWs. Methods: The study included 5576 HCWs from four Italian University Hospitals within a European Project, called ORCHESTRA. Socio-demographic and clinical data were collected retrospectively via online surveys from March 2020 to September 2022. Factors associated with SARS-CoV-2 infection during different pandemic periods were evaluated by a multinomial logistic regression model. Results were expressed as Relative Risk Ratios (RRR). Results: The cumulative incidence was 46.2%. The highest incidence period was the Omicron phase (OVP) (69.7%). The extra-professional source was the most reported (34.3%), followed by the occupational (26.8%). However, in almost 40%, the source was undetected. The RRR for occupational exposures was 0.39 (95% CI 0.25–0.61) during the Pre-Omicron variant Period (POP) and even lower (0.22, 95% CI 0.16–0.29) in the OVP, as compared to extra-professional exposures, using the Pre-Vaccination Period (PVP) as reference. Conclusions: The dominant source of infection among HCWs changed over time. While occupational contacts were more frequent during PVP, it significantly waned over the subsequent pandemic phases. Implementing procedures and guidelines to prevent infection, even outside the workplace during pandemics, would reduce the spread of infection among HCWs and patients. Full article

Oral manifestations are common in people living with HIV (PLWH) and may affect oral health-related quality of life (OHRQoL), while data from Greece remain limited. This multicenter prospective cohort study evaluated oral health status and OHRQoL among PLWH and explored associations with antiretroviral therapy (ART) and clinical factors. Overall, 370 PLWH from seven referral centers were included. Participants underwent oral examination, with oral hygiene assessed using the Simplified Oral Hygiene Index (OHI-S), and completed the Oral Health Impact Profile-14 (OHIP-14). Statistical analyses were performed using IBM SPSS Statistics v29.0, while multinomial and binary logistic regression identified predictors of oral hygiene status and OHRQoL, respectively. Most participants were male (76.5%), had CD4 counts ≥ 200 cells/μL (95.4%), and were receiving ART (98.6%). Annual dental check-ups, daily tooth brushing, mouthwash use, and dental floss use were reported by 54.1%, 69.5%, 31.9%, and 23.8%, respectively. The median OHI-S score was 2.0 (IQR:1.5–2.7), with 16.9% having poor OHI-S; the median OHIP-14 score was 11 (IQR: 7–15), with 64.4% reporting poor OHRQoL. Male sex was associated with lower odds of poor OHRQoL (OR = 0.377; p = 0.006), whereas ART regimen independently predicted poor OHRQoL. These findings support patient-centered oral healthcare within HIV care. Full article