Browse Articles
Discover research articles across all indexed journals
Discovery of a new mitophagy-related gene signature for predicting the outlook and immunotherapy in triple-negative breast cancer
Explicitly unbiased large language models still form biased associations
Large language models (LLMs) can pass explicit social bias tests but still harbor implicit biases, similar to humans who endorse egalitarian beliefs yet exhibit subtle biases. Measuring such implicit biases can be a challenge: As LLMs become increasingly proprietary, it may not be possible to access their embeddings and apply existing bias measures; furthermore, implicit biases are primarily a concern if they affect the actual decisions that these systems make. We address both challenges by introducing two measures: LLM Word Association Test, a prompt-based method for revealing implicit bias; and LLM Relative Decision Test, a strategy to detect subtle discrimination in contextual decisions. Both measures are based on psychological research: LLM Word Association Test adapts the Implicit Association Test, widely used to study the automatic associations between concepts held in human minds; and LLM Relative Decision Test operationalizes psychological results indicating that relative evaluations between two candidates, not absolute evaluations assessing each independently, are more diagnostic of implicit biases. Using these measures, we found pervasive stereotype biases mirroring those in society in 8 value-aligned models across 4 social categories (race, gender, religion, health) in 21 stereotypes (such as race and criminality, race and weapons, gender and science, age and negativity). These prompt-based measures draw from psychology’s long history of research into measuring stereotypes based on purely observable behavior; they expose nuanced biases in proprietary value-aligned LLMs that appear unbiased according to standard benchmarks.
Developing a dynamic combined power quality index for assessing the performance of a nuclear facility
Abstract Studying and evaluating the power quality (PQ) of an electrical network for nuclear installation is an important issue and a hot research topic for guaranteeing reliable and safe operation of sensitive electrical loads during this type of installation. As several PQ phenomena determine the overall PQ performance, analyzing PQ signals for evaluating the overall PQ is one of the major challenges for researchers in this field. Technically, voltage imbalance, current imbalance, voltage harmonic distortion, current harmonic distortion, and the power factor are five important PQ phenomena that judge the overall PQ performance of an electrical system. Multicriteria decision-making (MCDM) is used here as a methodology to identify a weighting for each PQ phenomenon. This paper proposes a power quality evaluation method for a nuclear research reactor (NRR) electrical network based on two MCDM. Methods the analytic hierarchy process (AHP) and criterion importance through inter-criteria correlation (CRITIC). A MATLAB/Simulink model for the NRR electrical system is presented, and then its validity and credibility are verified via measurements. In this study, different abnormal conditions are simulated in the NRR network to generate power quality disturbances, including a three-phase nonlinear load to simulate harmonics, an unbalanced load to simulate unbalance, and an inductive load to simulate the change in the power factor. The effectiveness and robustness of the proposed methodology are demonstrated through these different case studies. The results show that the obtained CPQI based on the dynamic weight approach allows for more accurate evaluations by adjusting the importance of various PQ phenomena depending on operational conditions and priorities. The main contribution of this paper is that a single compound power quality index (CPQI) was developed based on both the dynamic weights obtained from AHP-CRITIC methods and the results of the five PQ phenomena obtained under different abnormal conditions, considering the threshold level for each of these PQ phenomena. The analysis of the obtained results shows that this method accurately evaluates the overall PQ performance.
Manipulating hydrogenation pathways enables economically viable electrocatalytic aldehyde-to-alcohol valorization
Electrocatalytic reduction (ECR) of furfural represents a sustainable route for biomass valorization. Unfortunately, traditional Cu-catalyzed ECR suffers from diversified product distribution and industrial-incompatible production rates, mainly caused by the intricate mechanism−performance relationship. Here, we manipulate hydrogenation pathways on Cu by introducing ceria as an auxiliary component, which enables the mechanism switching from proton-coupled electron transfer to electrochemical hydrogen-atom transfer (HAT) and thus high-speed furfural-to-furfuryl alcohol electroconversion. Theoretical and kinetic analyses show that oxygen-vacancy-rich ceria delivers an efficient formation−diffusion−hydrogenation chain of H* by diminishing H* adsorption. Spectroscopic characterizations indicate that Cu/ceria interfacial perimeter enriches the local furfural, synergistically lowering the barrier of the rate-determining HAT step across the perimeter. Our Cu/ceria catalyst realizes high-rate HAT-dominated ECR for electrosynthesis of single-product furfuryl alcohol, achieving a high production rate of 19.1 ± 0.4 mol h −1 m −2 and a Faradaic efficiency of 97 ± 1% at an economically viable partial current density of over 0.1 A cm −2 . Our results demonstrate a highly efficient route for biofeedstock valorization with enhanced techno-economic feasibility.
The investigation of nonlinear vibration on metaconcrete single aggregate system and aggregate optimal design
A structural atlas of death domain fold proteins reveals their versatile roles in biology and function
Death domain fold (DDF) superfamily proteins are critically important players in pathways of cell death and inflammation. DDFs are often essential scaffolding domains in receptors, adaptors, or effectors of these pathways by mediating homo- and hetero-oligomerization including helical filament assembly. At the downstream ends of these pathways, effector oligomerization by DDFs brings the enzyme domains into proximity for their dimerization and activation. Hundreds of structures of these domains have been solved. However, a comprehensive understanding of DDFs is lacking. In this article, we report the curation of a DDF structural atlas as a public website (deathdomain.org) and deduce the common and distinct principles of DDF-mediated oligomerization among the four families (death domain or DD, death effector domain or DED, caspase recruitment domain or CARD, and pyrin domain or PYD). We further annotate DDFs genome-wide based on AlphaFold-predicted models and protein sequences. These studies reveal mechanistic rules for this widely distributed domain superfamily.
Neuroleptics used in critical COVID associated with moderate-severe dyspnea after hospital discharge
Characterization of jackfruit Artocarpus heterophyllus Lam. for economic traits in dry zones of Karnataka, India
<i>Enterobacter hormaechei</i> replaces virulence with carbapenem resistance via porin loss
Pathogenic Enterobacter species are of increasing clinical concern due to the multidrug-resistant nature of these bacteria, including resistance to carbapenem antibiotics. Our understanding of Enterobacter virulence is limited, hindering the development of new prophylactics and therapeutics targeting infections caused by Enterobacter species. In this study, we assessed the virulence of contemporary clinical Enterobacter hormaechei isolates in a mouse model of intraperitoneal infection and used comparative genomics to identify genes promoting virulence. Through mutagenesis and complementation studies, we found two porin-encoding genes, ompC and ompD , to be required for E. hormaechei virulence. These porins imported clinically relevant carbapenems into the bacteria, and thus loss of OmpC and OmpD desensitized E. hormaechei to the antibiotics. Our genomic analyses suggest porin-related genes are frequently mutated in E. hormaechei , perhaps due to the selective pressure of antibiotic therapy during infection. Despite the importance of OmpC and OmpD during infection of immunocompetent hosts, we found the two porins to be dispensable for virulence in a neutropenic mouse model. Moreover, porin loss provided a fitness advantage during carbapenem treatment in an ex vivo human whole blood model of bacteremia. Our data provide experimental evidence of pathogenic Enterobacter species gaining antibiotic resistance via loss of porins and argue antibiotic therapy during infection of immunocompromised patients is a conducive environment for the selection of porin mutations enhancing the multidrug-resistant profile of these pathogens.
Efficacy of Naringenin against aging and degeneration of nucleus pulposus cells through IGFBP3 inhibition
Temporal autocorrelation is predictive of age—An extensive MEG time-series analysis
Understanding the evolving dynamics of the brain throughout life is pivotal for anticipating and evaluating individual health. While previous research has described age effects on spectral properties of neural signals, it remains unclear which ones are most indicative of age-related processes. This study addresses this gap by analyzing resting-state data obtained from magnetoencephalography (MEG) in 350 adults (18 to 88 y). We employed advanced time-series analysis at the brain region level and machine learning to predict age. While traditional spectral features achieved low to moderate accuracy, over a hundred time-series features proved superior. Notably, temporal autocorrelation (AC) emerged as the most robust predictor of age. Distinct patterns of AC within the visual and temporal cortex were most informative, offering a versatile measure of age-related signal changes for comprehensive health assessments based on brain activity.
Detecting severe coronary artery stenosis in T2DM patients with NAFLD using cardiac fat radiomics-based machine learning
A novel quinone biosynthetic pathway illuminates the evolution of aerobic metabolism
The dominant organisms in modern oxic ecosystems rely on respiratory quinones with high redox potential (HPQs) for electron transport in aerobic respiration and photosynthesis. The diversification of quinones, from low redox potential (LPQ) in anaerobes to HPQs in aerobes, is assumed to have followed Earth’s surface oxygenation ~2.3 billion years ago. However, the evolutionary origins of HPQs remain unresolved. Here, we characterize the structure and biosynthetic pathway of an ancestral HPQ, methyl-plastoquinone (mPQ), that is unique to bacteria of the phylum Nitrospirota . mPQ is structurally related to the two previously known HPQs, plastoquinone from Cyanobacteriota /chloroplasts and ubiquinone from Pseudomonadota /mitochondria, respectively. We demonstrate a common origin of the three HPQ biosynthetic pathways that predates the emergence of Nitrospirota , Cyanobacteriota , and Pseudomonadota . An ancestral HPQ biosynthetic pathway evolved ≥ 3.4 billion years ago in an extinct lineage and was laterally transferred to these three phyla ~2.5 to 3.2 billion years ago. We show that Cyanobacteriota and Pseudomonadota were ancestrally aerobic and thus propose that aerobic metabolism using HPQs significantly predates Earth’s surface oxygenation. Two of the three HPQ pathways were later obtained by eukaryotes through endosymbiosis forming chloroplasts and mitochondria, enabling their rise to dominance in modern oxic ecosystems.
Posterior approach unilateral laminectomy, debridement, and preshaped titanium mesh bone grafting with internal fixation for treatment of lumbar tuberculosis
Poly(ADP-ribose) binding sites on collagen I fibrils for nucleating intrafibrillar bone mineral
Bone calcification is essential for vertebrate life. The mechanism by which mineral ions are transported into collagen fibrils to induce intrafibrillar mineral formation requires a calcium binding biopolymer that also has highly selective binding to the collagen fibril hole zones where intrafibrillar calcification begins, over other bone extracellular matrix components. Poly(ADP-ribose) (PAR) has been shown to be a candidate biopolymer for this process and we show here that PAR has high affinity, highly conserved binding sites in the collagen type I C-terminal telopeptides. The identification of these PAR–collagen binding sites gives insights into the chemical mechanisms underlying bone calcification and possible mechanisms behind pathologies where there is dysfunctional bone calcification.
The association of chronic pain, painkiller use, and potential mediators with liver fat content
The impacts of power transmission and transformation projects on ecological corridors and landscape connectivity: a case study of Shandong province, China
Learning-based inference of longitudinal image changes: Applications in embryo development, wound healing, and aging brain
Longitudinal imaging data are routinely acquired for health studies and patient monitoring. A central goal in longitudinal studies is tracking relevant change over time. Traditional methods remove nuisance variation with custom pipelines to focus on significant changes. In this work, we present a machine learning–based method that automatically ignores irrelevant changes and extracts the time-varying signal of interest. Our method, called Learning-based Inference of Longitudinal imAge Changes (LILAC), performs a pairwise comparison of longitudinal images in order to make a temporal difference prediction. LILAC employs a convolutional Siamese architecture to extract feature pairs, followed by subtraction and a bias-free fully connected layer to learn meaningful temporal image differences. We first showcase LILAC’s ability to capture key longitudinal changes by simply training it to predict the temporal ordering of images. In our experiments, temporal ordering accuracy exceeded 0.98, and predicted time differences were strongly correlated with actual changes in relevant variables (Pearson Correlation Coefficient r = 0.911 with embryo phase change, and r = 0.875 with time interval in wound healing). Next, we trained LILAC to explicitly predict specific targets, such as the change in clinical scores in patients with mild cognitive impairment. LILAC models achieved over a 40% reduction in root mean square error compared to baseline methods. Our empirical results demonstrate that LILAC effectively localizes and quantifies relevant individual-level changes in longitudinal imaging data, offering valuable insights for studying temporal mechanisms or guiding clinical decisions.