Search Results (21,740)

Volatile organic compounds (VOCs) are ubiquitous pollutants, and exposure from in utero through childhood may impair neurodevelopment. However, compound-specific risks remain unclear. This systematic review and meta-analysis examined associations between VOC exposure and child neurodevelopmental outcomes. A systematic search of PubMed, Scopus, and Embase was conducted until August 2025, yielding 1213 records. Quality assessment was performed using the Newcastle–Ottawa Scale and risk of bias using the ROBINS-E tool. Pooled odds ratios (ORs) were calculated using random-effects models, with heterogeneity evaluated via I² statistics. Subgroup analyses explored for study design, exposure timing, and country income level. Twenty-eight studies were included in the final analysis. Of the 18 VOCs analyzed, five compounds, propionaldehyde (pooled OR = 1.84; 95% CI 1.19–2.49), styrene (pooled OR = 1.69; 95% CI 1.30–2.21), vinyl chloride (pooled OR = 1.53; 95% CI 1.24–1.89), acrolein (pooled OR = 1.48; 95% CI 1.08–2.04), and trichloroethylene (OR = 1.21; 95% CI 1.04–1.41), demonstrated statistically significant adverse associations with neurodevelopment. Benzene showed borderline significance. Heterogeneity ranged from 0–47%. Subgroup analyses identified significant effect modification for 1,3-butadiene by study design and exposure timing and higher pooled estimates for ethylbenzene in high-income countries. Full article

Softmax is a step in transformer computation during which the internal buffer size grows rapidly because of the use of the exponential function. Softmax is a fundamental yet computationally expensive operation in vision transformer attention, posing significant challenges for deployment on resource-constrained FPGAs (Field Programmable Gate Arrays). Computational precision demands grow at the softmax stage in the attention pipeline mainly because of the use of the exponential function in the softmax computation. This paper proposes a low-precision softmax approximation that combines a truncated Maclaurin-series exponential with input-range clamping to enable efficient hardware realization without sacrificing reconstruction quality. By bounding extreme attention scores that contribute negligibly to final outputs, the proposed method mitigates the instability of low-order polynomial approximations while preserving their hardware efficiency. The approach is first validated in software using SwinIR (Image restoration using the SWIN Transformer) super resolution to ensure reconstruction fidelity and is then analyzed for FPGA deployment. SWINIR is a multi-stage version of other transformers like Deit and Vit, making it a preferred option for testing the reconstruction fidelity of the change for transformers. Experimental results demonstrate that the proposed fourth-order clamped approximation achieves near-reference performance, incurring only 0.15 dB PSNR and 0.0059 SSIM degradation on SwinIR-M, while significantly reducing precision and memory requirements. For the large-sized SWINIR model (SWINIR-L), a PSNR increase with a less than 0.01 SSIM loss is observed, further highlighting the insignificance of extreme values as model size gets bigger. A Horner-form reformulation further improves hardware efficiency by limiting intermediate precision growth. Overall, this work presents a reconstruction-aware and hardware-friendly softmax reformulation that enables practical deployment of vision transformers on small FPGA platforms. This work also uses this contribution to improve the performance of the ViTA accelerator design. We also add bias initialization and a PE loop bound runtime variable to the existing ViTA accelerator design. Full article

Application Tracking Systems (ATSs) have evolved significantly since their inception in 1996, transitioning from simple resumérepositories to AI-driven tools with advanced capabilities. While these developments have improved recruitment efficiency, they have also raised important ethical, organizational, and human-rights-related concerns. Bias in machine learning (ML) training data, opaque decision criteria, and excessive reliance on automated judgment may contribute to unfair treatment, reduced transparency, and limited human oversight in hiring processes. This study addresses these challenges by proposing a human-centered approach to ATS-supported recruitment based on a set of Humanization Services. Using a Design Science Research approach, three main artifacts were developed: a Job Requirements Validation Module, a Bias Trigger Removal Module, and a blockchain-supported dual-authorization mechanism for vacancy approval, which requires digital signatures from qualified professionals to approve job postings, ensuring that there are humans that assume responsibility. These components are intended to improve job posting quality, reduce bias-conducive information in applicant data, and strengthen accountability in recruitment workflows. The evaluation provides initial empirical support for the operational feasibility of the proposed approach under the tested conditions. The study therefore contributes a practical and theoretically grounded step toward more transparent, accountable, and human-centered AI-supported recruitment. Full article

The DNA methyltransferases inhibitor 5-azacytidine (5AzC), clinically used to treat hematopoietic malignancies, can elevate genomic mutational burden, raising safety concerns. To define the epigenetic specificity and mutagenic consequences of 5AzC, we performed multi-omics analyses in Neurospora crassa. Our data showed that 5AzC caused a non-selective, genome-wide reduction in both 5-methylcytosine (5mC; ~50% decrease) and the heterochromatin mark H3K9me3 (~65% decrease), indicating broad off-target demethylation that may transiently benefit therapy yet compromise genome stability. Whole-genome sequencing (WGS) revealed a ~290-fold increase in mutation rate under 5AzC, with a pronounced C->G transversion bias, a spectrum typically associated with higher functional burden. Strikingly, 5AzC-induced mutations were enriched in H3K9me3-marked domains, particularly pericentromeric regions characterized by low 5mC but high H3K9me3. Genetic analyses showed that the loss of DNA methyltransferase DIM-2 reduced 5AzC-induced mutations by ~64%, while individual or combined knockout of the histone methyltransferase DIM-5 with DIM-2 led to an 85% reduction. Thus, mutagenesis was markedly amplified by DIM-2 and DIM-5, with DIM-2 activity dependent on DIM-5. Collectively, DIM-2 and DIM-5 accounted for nearly all A/T-site and ~80% of G/C-site mutations. These results reveal that 5AzC drives genome-wide loss of 5mC and H3K9me3, with mutagenesis preferentially targeting H3K9me3-enriched regions via DIM-2 and DIM-5. This work clarifies a mechanistic basis for 5AzC-associated genomic risk and highlights strategies for next-generation epigenetic therapies that preserve heterochromatin integrity while minimizing mutational load. Full article

Semiconductor lasers have been widely employed in chaos-based information processing due to their ability to generate enhanced chaotic bandwidths. In this study, we investigate broadband polarization chaos in optically injected QD spin-VCSELs and their ability to act as high-speed physical entropy sources for random number generation (RNG). We achieve chaotic bandwidths approaching 50 GHz per polarization mode using elliptical injection. With optimized conditions and post-processing, we demonstrate RNG at rates of up to 240 Gb/s. The quality of the generated random sequences is evaluated using multiple statistical metrics, including entropy estimation based on the NIST SP800-90B framework, uniqueness analysis using Hamming distance, and bias assessment through autocorrelation and histogram analysis. In addition, the influence of different polarization injection schemes on randomness is examined using the NIST SP800-22 statistical test suite. These results highlight the potential of QD spin-VCSELs as compact and ultrafast sources for RNG in secure communication systems. Full article

This paper presents a nanowatt-scale operational transconductance amplifier (OTA) and an electronically tunable third-order low-pass filter (LPF) designed for energy-constrained biomedical signal conditioning. The circuits are implemented in a 65 nm CMOS process and verified through comprehensive schematic-level simulations. Biased in the deep subthreshold region at 1 nA, the OTA achieves a 50 dB low-frequency gain, a 225 Hz unity-gain bandwidth at 10 pF load capacitance and an input-referred noise floor of 1.55 μV/√Hz, with a total power consumption of only 1.75 nW. The integrated third-order LPF provides a wide tuning range (37–668 Hz) via bias current modulation, exhibiting excellent linearity with a THD of 0.059% and a 65.3 dB dynamic range. Monte Carlo and PVT corner analyses demonstrate the design’s theoretical robustness against process variations and environmental fluctuations. ECG signal simulations validate the circuit’s effectiveness in suppressing high-frequency artifacts while preserving morphological integrity, providing a proof-of-concept for ultra-low-power wearable healthcare architectures. Full article

Pulsed electric field (PEF) extraction studies often report yield gains but less consistently translate them into purification-relevant metrics. This study re-analyzed a locked comparability tier of 56 solvent-matched control (CTR)-PEF contrasts from five leaf-biomass studies to map total phenolic content (TPC) uplift onto first-order downstream duty proxies. The tier was not designed as a field-wide survey, and audit logs plus dependence-aware analyses were used to reduce author bias. Across the 56 pairs, median ΔTPC% was 26.7% (IQR 12.1–35.0), with positive values in 55 of 56 contrasts. The strongest gains were concentrated at low specific energy (Wspec ≤ 0.15 kJ kg⁻¹ treated biomass) and 0–25% ethanol within this dataset. PLRI translated these uplifts into first-order processed-volume scaling at fixed capture criteria, and Pareto screening identified PLRI values of 1.18–1.73, corresponding to approximately 15–42% lower processed volume under the stated assumptions. Marker-level data were limited to two matrices (three paired observations), but SSF values > 1 (1.06–2.89) were consistent with possible composition steering. No new experiments were performed, and ballast, fouling, and downstream performance were not measured; conclusions are therefore limited to the operating envelope and are intended for screening, reporting standardization, and subsequent purification validation. Full article

Background: Nutrition and sleep are critical determinants of athletic performance and recovery. Direct comparative research between endurance and strength–power athletes remains limited. This study aimed to evaluate and compare nutritional knowledge, dietary habits, sleep quality, and Body Mass Index between ultramarathon runners and American football players, as well as to explore independent predictors of sleep quality. Methods: A cross-sectional study was conducted among 231 male athletes. To address group size disparity and mitigate statistical bias, a random undersampling technique was applied to create a balanced cohort of 86 athletes comprising 43 ultramarathon runners and 43 American football players. Nutritional parameters were assessed using the Kom-PAN questionnaire. Sleep quality was evaluated using the Pittsburgh Sleep Quality Index. Between-group comparisons were performed using the Mann–Whitney U test with False Discovery Rate correction. An integrated multiple regression model was constructed to identify predictors of global sleep quality. Results: Ultramarathon runners demonstrated significantly better overall sleep quality (p = 0.026) and higher nutritional knowledge (p < 0.001) compared to American football players. Differences in adherence to pro-healthy and non-healthy dietary patterns were not statistically significant after False Discovery Rate correction. The integrated multiple regression model revealed that the athletic discipline was the primary independent predictor of global sleep quality (p = 0.001), while dietary variables did not exhibit a significant independent effect. Furthermore, higher Body Mass Index was independently associated with better sleep scores within the multivariate model (p = 0.008). Conclusions: Significant sport-specific differences exist in BMI, nutritional knowledge, and sleep quality. Global sleep quality appears to be primarily associated with the specific physiological and environmental demands of the athletic discipline rather than individual dietary factors, which were not independently significant in the multivariable model. These findings suggest that recovery strategies in strength–power athletes may require a broader, multifactorial approach beyond nutritional education alone. Full article

We investigate the effects of post-annealing in oxygen (O₂) and nitrogen (N₂) on tin-doped vanadium oxide (V_xSn_yO_z) films for microbolometer applications. The films were deposited using magnetron sputtering in an Ar:O₂ environment. We demonstrate that low Sn doping combined with N₂ post-annealing provides an effective approach to optimize the temperature coefficient of resistance (TCR), resistivity, and 1/f-noise. Compared to undoped VO_x, V_xSn_yO_z films exhibit an enhanced TCR, moderate resistivity, and reduced 1/f-noise. The 135 nm thick V_0.46Sn_0.03O_0.51 film after post-annealing in N₂ shows a TCR of −4.08%/K and a resistivity of 7.3 × 10⁻² Ω⋅cm at 300 K, an absorptance of 63–68% in the 900–2500 nm wavelength range, and low noise voltage power spectral density (1.77 × 10⁻¹⁶ V²/Hz at 100 Hz under 0.3μA bias current). These results indicate that Sn-doped VO_x films are promising sensing materials for microbolometer applications. Full article

A ternary CdS–MoS₂–rGO photocathode was developed to enhance visible light-driven hydrogen evolution through interfacial heterostructure engineering. The composite was fabricated via a solution-based deposition method followed by thermal conversion, resulting in crystalline CdS and MoS₂ phases that were uniformly integrated within a conductive reduced graphene oxide (rGO) framework. Structural and surface analyses (XRD and XPS) confirmed the coexistence of Cd²⁺, Mo⁴⁺, and S²⁻ chemical states without detectable secondary phases. Photoelectrochemical measurements revealed that the ternary architecture significantly improves charge separation efficiency and interfacial charge-transfer kinetics compared to binary and single-component films. The CdS–MoS₂–rGO photocathode exhibited the highest photocurrent density, reduced charge-transfer resistance, and favorable Tafel slope under visible-light irradiation (0.25 sun, neutral electrolyte). Gas chromatography measurements verified that these electrochemical enhancements translate into increased hydrogen production rates, following the trend: CdS–MoS₂–rGO > CdS–rGO > MoS₂–rGO >> rGO. Applied bias photon-to-current efficiency (ABPE) analysis further confirmed improved photon utilization efficiency in the ternary system. The enhanced performance is attributed to synergistic integration of CdS (light harvesting), rGO (rapid electron transport), and MoS₂ (catalytic edge sites), which suppresses recombination and accelerates proton reduction kinetics. These findings demonstrate that rational multi-component heterostructure design is an effective strategy for improving hydrogen evolution rate under mild operating conditions. Full article

This study systematically examines the intersection of emerging digital technologies and organizational accountability within the sustainability domain using the TCCM framework. Guided by the SPAR-4-SLR protocol, a final corpus of 67 high-impact peer-reviewed articles was analyzed to synthesize current knowledge and identify structural gaps in governance architectures. Findings indicate that traditional human-led narrative disclosures are increasingly supplemented or replaced by technology-embedded verification systems offering real-time data granularity. The analysis shows that while the field is largely grounded in Stakeholder Theory and the Resource-Based View, mid-range theorizing is needed to address algorithmic bias and the gap between technological capabilities and accountability practices. Empirical evidence is concentrated in Europe and East Asia, exposing a digital divide that limits the applicability of findings to resource-constrained enterprises. The study provides a conceptual synthesis of how AI, blockchain, and IoT reshape transparency, highlighting the need for governance approaches that prioritize ethical oversight, decentralized validation, and substantive rather than symbolic compliance. Full article

Light hydrocarbons in shale oil readily volatilize during conventional coring, surface handling, storage, and laboratory preparation. The resulting evaporative loss causes systematic underestimation of Rock-Eval S₁ peak (free hydrocarbons measured by programmed pyrolysis), which can bias oil-bearing evaluation, sweet-spot delineation, and resource assessment. Here we investigate Chang 7₃ lacustrine shale oil in the Ordos Basin (China) using frozen cores recovered by pressure-retained coring from four wells. Time-series Rock-Eval pyrolysis and thermal desorption–gas chromatography (TD–GC) were used to quantify the magnitude, temporal evolution, and practical equilibrium time of light-hydrocarbon loss and to establish a practical restoration model. S₁ decreases with storage time and exhibits a clear two-stage behavior. Shale shows a rapid-loss stage during 0–90 days, followed by a practical equilibrium stage after 90 days (S₁ change less than 5%). Sandstone interbeds lose light hydrocarbons faster and more completely, reaching practical equilibrium after 60 days. TD–GC indicates that the lost fraction is dominated by n-alkane components lighter than C₁₃, with gaseous hydrocarbons showing the greatest depletion; medium and heavy fractions decrease modestly. Loss is coupled with progressive desorption from kerogen and clays, leading to enrichment of heavier components in the residual free hydrocarbons and a shift of the modal carbon number toward higher values. At the shale equilibrium time, total organic carbon (TOC) and vitrinite reflectance (R_o) jointly control the restoration factor K. We propose a two-parameter restoration model: K = (0.4024·ln (TOC) + 0.821)·(0.652·R_o + 0.4292). Applying the model to more than 50 conventionally cored wells reveals that the Qingyang–Zhengning area in the southwestern basin is the principal enrichment zone of original free hydrocarbons, followed by the Jiyuan area in the north and the Huachi area in the central basin, whereas the eastern basin is relatively depleted. The workflow provides a robust and transferable approach for correcting S₁ and improving shale-oil evaluation in lacustrine systems. Full article

Single-chip integration technology for multifunctional sensors has become an important development direction due to its low power consumption and versatile functionality. However, the fabrication compatibility between different sensing components remains a key challenge for high-performance integrated sensors, often leading to complex processes and increased costs. This work presents a microfabrication-compatible photo-magnetic integrated sensor chip based on micro–nano processing methods. The integrated sensor chip includes giant magnetoresistance (GMR) and a light-dependent resistor (LDR). The fabrication process was based on standard MEMS fabrication with compatibility and cost-effectiveness. The experimental results demonstrated that the chip can simultaneously realize both optical and magnetic detection with magnetic field sensitivity of 3.74 mV/Oe and photodetection sensitivity of 0.79 $\mathsf{\mu }$A/($\mathsf{\mu }$W/cm²) at a 5 V bias. The integrated sensor features high-sensitivity magnetic performance and weak-light detection capability, with promising application in robotics and advanced manufacturing fields. Full article

Large Language Models (LLMs) are increasingly deployed as automated annotators in semantic multimedia systems, yet their reliability varies significantly across architectures. This study extends prior cross-model evaluations by benchmarking ChatGPT-5, Qwen-3, and Gemini-3-flash against human expert annotations using the HRAST hotel review dataset. We adopt a bias-by-design framework to analyze systematic divergences in sentiment, topic, and aspect labeling across real and synthetic data, while investigating the moderating effects of annotation mode. Findings reveal model-contingent polarity bias: ChatGPT-5 exhibits a pronounced neutrality bias, while Qwen-3 and Gemini-3-flash align more closely with human polarization. Agreement is substantial for concrete topics but diverges on abstract evaluative dimensions. Synthetic data consistently inflates reliability metrics while masking ambiguity. These findings highlight that annotation bias is structurally embedded in model design choices and operational conditions. Cross-architectural triangulation and mode-aware deployment strategies are recommended for robust semantic multimedia system development. Full article

Background: Hematological patients have a high risk of developing severe COVID-19 (37%). Most mRNA vaccine trials in hematological patients showed a low immunogenicity after two doses, while long-term data are scarce. Methods: In this monocentric retrospective observational study, we evaluated humoral and T cell-mediated immune responses in 230 hematological patients after three doses of the Pfizer-BioNTech mRNA COVID-19 vaccine. Patients were stratified by age, disease type/state, prior COVID-19 infection, and treatment status and regimens (anti-CD20 monoclonal antibodies, BTK and BCL-2 inhibitors, and treatment line). Antibody titer to SARS-CoV-2 was assessed by electrochemiluminescence immunoassay and T cell response by QuantiFERON interferon-γ release assay (IGRA). Data were analyzed using univariate (Fisher’s exact test) and Firth’s bias-reduced penalized-likelihood logistic regression. Results: A robust humoral response was observed with 91.55% of patients developing anti-spike antibodies (GMT 988.83 U/mL). Anti-CD20-bendamustine treatment was associated with a significantly lower antibody positivity compared to untreated subjects. Prior COVID-19 infection significantly boosted both antibody positivity (95.9% vs. 85.2%) and GMT (847.02 U/mL vs. 258.79 U/mL). Conversely, T cell response was suboptimal (36.1% positive), particularly in anti-CD20-bendamustine-treated and multi-treated patients (27.1%), but highest in those treated with BTK inhibitors (50%). Multivariable logistic regression analysis linked multiple treatments to lower T cell response. Following vaccination, 29.1% of patients contracted SARS-CoV-2, but only 0.89% developed severe COVID-19. Conclusions: Three doses of mRNA vaccine elicit a strong humoral but a low T cell response, as detected by IGRA, in hematological patients. These findings underscore the importance of completing vaccination before initiating immunosuppressive therapies. Full article