Browse Articles
Discover research articles across all indexed journals
Graph neural networks for networked analysis of gestational diabetes risk factors: a multi method framework
Abstract Gestational diabetes mellitus, often known as GDM, is a major health issue that causes complications for mothers and requires patient data prediction models that are complex and variable. The research in question makes use of graph-based learning in order to investigate the ways in which genetic, biochemical, and demographic elements interact in a variety of different contexts. Through the use of nodes to represent patients and lines to represent the things that they share in common, the framework illustrates how the aforementioned elements influence the likelihood of illness. Graph neural networks are utilized for the process, while BioBERT embeddings are utilized for the management of unstructured clinical notes. Graph neural networks are utilized for organized clinical notes. Because of this alignment, healthcare processes are placed in the context in which they should be, rather than being taken out of context while they are being carried out. The graph architecture used in BioBERT incorporates semantic patterns derived from medical information into a relational structure that illustrates the degree to which patients are similar to one another. After being evaluated on a substantial clinical dataset, the proposed method is able to make more accurate and readable predictions than the baseline models. The results of this study indicate that the utilization of graph architecture with both organized and unstructured data can assist in the discovery of novel approaches to the treatment of GDM that go beyond performance sets. According to the findings of the study, machine learning needs to be modified so that it can be used with healthcare applications.
Hybrid physics-informed machine learning framework for calibration-free degradation prediction of lithium-ion batteries
Correction: Erysipelothrix rhusiopathiae clone reemergence in association with a multi-year mass mortality event in high Arctic muskoxen (Ovibos moschatus)
Daily briefing: Trial to ‘de-age’ cells treats first person
Detrimental influence of high neuroticism on visuospatial selective attention: behavioral and ERP evidence
Retraction Note: Predicting the compressive strength of polymer-infused bricks: A machine learning approach with SHAP interpretability
An information-theoretic evaluation framework for CNN–LSTM-based Alzheimer’s disease classification from structural MRI
Abstract Early detection of Alzheimer’s disease (AD) is important because of its progressive impact on cognitive function. This study presents a CNN–LSTM-based framework for three-class AD classification from structural MRI, with the primary contribution being a post-hoc information-theoretic evaluation strategy rather than a new network architecture. Experiments were conducted using 827 ADNI subjects, including normal controls (NC: 340), mild cognitive impairment (MCI: 307), and AD (180). To mitigate data scarcity and improve training diversity, GAN-based augmentation was applied only to the training data, while validation and test subjects were kept separate. In addition to conventional metrics, trained models were evaluated using Renyi mutual information, Renyi divergence, and Henze–Penrose divergence to quantify information preservation, representation stability, and distributional alignment. Under a subject-level evaluation protocol, the CNN–LSTM model achieved 96.7% accuracy and outperformed evaluated benchmark architectures under the same protocol. The information-theoretic measures provided complementary evidence for comparing model behavior beyond accuracy, particularly regarding information retention and output-distribution alignment. Overall, the findings suggest that post-hoc information-theoretic analysis can support more transparent assessment of MRI-based AD classification models. However, external validation on independent multi-center datasets is required before clinical deployment can be considered.
Cryo-EM structures of Měnglà virus GP reveal combined Ebola- and Marburg-like epitope masking strategies for antibody evasion
Ebola virus (EBOV) and Marburg virus (MARV) are highly lethal filoviruses that cause severe hemorrhagic fever in humans. A recently identified bat-borne filovirus, Měnglà virus (MLAV), uses the same NPC1 receptor as EBOV and MARV, raising concerns about its potential cross-species transmission. Here, we report cryo-EM structures of the MLAV surface glycoprotein (GP) in its unbound form and in complex with the MARV-neutralizing antibody MR191. MLAV GP exhibits distinctive structural features in the Wing and heptad repeat 1D (HR1D) regions, retains a visible Cap structure even after protease treatment, and contains a MARV GP-like α2 helix. MR191, a broadly neutralizing marburgvirus antibody that targets the conserved NPC1 receptor-binding pocket in MLAV GP, nonetheless exhibits impaired neutralizing activity, likely due to shielding by the MLAV Cap. In addition, the MLAV mucin-like domain, α2 helix, and HR1A region hinder binding by representative broadly neutralizing ebolavirus antibodies targeting the GP-waist, including 6D6, CA45, ADI-15878, and ADI-15946. Together, these results provide the first structural insights into MLAV GP and identify immune evasion driven by structural and sequence divergence as a major challenge for pan-filovirus antibody development.
Development of an RFID sensor tag reading method in shielded areas based on an extended antenna
Enhancing environmental sustainability in Somalia through water conservation, pollution reduction, energy saving, and waste management efficiency
APOBEC2 deficiency disrupts hematopoietic lineage commitment, resulting in emergence of dual identity lymphocytes in mice and humans
APOBEC2 is a well-conserved member of the AID/APOBEC family of cytidine deaminases. Most members of the family catalyze the conversion of cytosine to uracil either in DNA or RNA, thereby acting as DNA mutators and/or RNA editors. APOBEC2 is the only family member that appears to be catalytically inactive. Instead, its ability to bind, but not deaminate DNA, has been co-opted into a transcription-factor-like functionality. APOBEC2 is highly expressed in skeletal muscle, where it functions to promote and maintain muscle identity by suppressing nonmuscle genes. APOBEC2 is also expressed in several cell types within the hematopoietic lineage. Here, we show that loss of APOBEC2 disrupts proper lymphoid lineage differentiation, resulting in the emergence of lymphoid cells expressing both TCRs and BCRs, as well as additional markers that define a mixed T and B cell identity in both mice with a depletion in Apobec2 gene and humans with mutations in it. We further show that these T/B cells present dual functionality. Finally, we elucidate the molecular mechanisms associated with APOBEC2 deficiency that led to disruption of cell fate determination. Overall, our results establish APOBEC2 as a terminal repressor of B cell fates within the T cell lineage, whose loss results in disease outcomes in mice and humans.
A unicellular relative links aggregative multicellularity to animal origins
Abstract How animals evolved complex multicellularity from their unicellular ancestors remains unanswered. Unicellular relatives of animals exhibit simple multicellularity through clonal division, formation of multinucleate coenocytes or aggregation 1 . Animal multicellularity may therefore have evolved from one (or a combination) of these behaviours. Aggregation has classically been dismissed as a means to complex multicellularity 2 . However, aggregation occurs in many extant animal cells and has also been recently described in three close unicellular relatives of animals (the choanoflagellates Salpingoeca rosetta and Choanoeca flexa , and the filasterean Capsaspora owczarzaki ) 3–5 . It is unclear whether aggregation in these species is derived or ancestral, and its relevance for animal origins remains unclear. Here, to fill this gap, we investigated whether an additional close unicellular relative of animals can undergo aggregation. We found that the marine free-living bacterivorous filasterean Ministeria vibrans 6 forms homogeneous aggregates with reproducible kinetics that have long-term stability, and that improved feeding and mating may be evolutionary drivers of this aggregation. Notably, we found that homologues of many animal multicellularity genes involved in cell adhesion, signalling and transcriptional regulation were deployed during the aggregation process, indicating that they may have been used for aggregation in the unicellular ancestors of animals before being co-opted into animal multicellular development. Thus, our results are consistent with aggregative multicellularity being key to the evolution of the multicellular animal genetic toolkit.
Contrasting cognitive control in the Simon and spatial Stroop tasks regarding their interference with the control of standing balance
Abstract The scientific understanding of any interaction between cognition and balance control is advanced by methods that capture event-related effects of cognitive processes on balance with high temporal resolution and precision. We developed such an approach to examine how cognitive conflict interferes with the control of body balance during upright standing. Participants stood on a force plate while performing two cognitive conflict paradigms: a Simon task, which according to Kornblum et al.’s dimensional overlap model 41 mainly induces spatial stimulus–response conflict during response selection, and a Spatial Stroop task, which additionally elicits a stimulus–stimulus conflict during stimulus encoding. By aligning force plate time series data to the onset events of target and response across all trials, we assessed the temporal dynamics of spatial congruency effects on force moment variability as a marker of balance control activity. Across both experimental cognitive tasks, we observed strong congruency effects in cognitive task performance, when considering trials after previous congruent trials. Further, incongruent trials were associated with systematic transient reductions in force moment variability along the mediolateral axis in balance control. These observations are in line with the assumption that the recruitment of cognitive processes for conflict resolution temporarily inhibits, suppresses, or postpones balance adjustments. Importantly, regarding the impact of cognitive interference on body balance, data confirm our previous observations using improved methods and demonstrate that reduction in balance control activity during resolution of cognitive conflict generalizes to a task with multiple conflict loci (Spatial Stroop task). Thereby, this extended range of conflict does not result in correspondingly stronger interference effects in balance control. From a theoretical perspective, the results align with predictive models of postural regulation and intermittent, event-driven accounts of balance control.
Mapping <i>CO</i> <sub>2</sub> fixation to two effective parameters: A framework toward data-informed species and model comparison
To improve crop yield and resilience, it is essential to identify the steps limiting C O 2 assimilation rate in plant leaves. The combined effect of multiple traits can be resolved by mechanistic models of the underlying diffusion, biochemistry, and geometry. Yet the widely used simple serial resistance models overlook tissue geometry, and detailed anatomical models are computationally heavy and rely on parameters that are difficult to measure. Here, we develop a framework for systematic species and model comparison, and find that the necessary level of model resolution is species-specific. We apply a minimal reaction–diffusion model and reduce C O 2 fixation in leaves to two key parameters. These parameters comprise a compact phase space in which three rate-limiting regimes emerge naturally: stomatal uptake, intercellular diffusion, and intracellular processes. Mapping diverse plant species into this phase space reveals: 1) dominant colimitations by stomatal and intracellular processes, 2) an equal partition between species that require spatially resolved leaf-scale models and species where intracellular models suffice. Taken together, we present a scalable path for interpreting complex trait data and bridging between models.
Ecosystem simulation: the software to platform leap
Abstract The transition from a software provider to a platform organizer is a pivotal transformation in the industrial software ecosystem (ISECO), enabling firms to fully leverage ecosystem dynamics, drive innovation, and achieve sustainable competitive advantages. Despite its growing significance, the strategic mechanisms underlying this transition remain underexplored. This study addresses this gap by employing an evolutionary game-theoretic approach to model the interplay among the government, the platform organizer, and the user in shaping ISECO’s trajectory across initial, mature, and ideal developmental stages. Through rigorous simulation analyses, we demonstrate that government subsidies and cost-sharing mechanisms significantly influence the evolutionary stability of the ecosystem. Furthermore, our findings reveal a critical threshold in cost-sharing strategies: a δ range of 0.2–0.4 that effectively incentivizes user adoption. When δ > 0.4, platform organizers may experience diminishing incentives to offer advanced services, potentially destabilizing the ISECO. This research advances the theoretical understanding of platform evolution by formalizing the dynamic, multi-agent interactions and threshold effects that drive this shift. It provides actionable insights for policymakers, software developers, and industrial strategists to design stage-specific subsidy withdrawal mechanisms and optimize cost-sharing structures without compromising profitability. The results underscore the necessity of dynamic, stage-specific interventions to facilitate a sustainable transition from software provision to platform organization.
Dynamic diversification of lignan metabolism in sesame via coordinated oxygenation and glucosylation across germination
Sesame ( Sesamum indicum ) seeds accumulate specialized lignans, including (+)-sesamin, (+)-sesamolin, and (+)-sesaminol triglucoside (SL-TG). Although lignan biosynthesis during seed development is well characterized—with SiCYP92B14 recognized as a (+)-sesamin-specific oxygenase—the molecular basis of the metabolic transition during germination, where lipophilic lignans are fully converted into glucosides, remained unclear. Herein, we identify a set of (+)-sesamin oxygenases, SiCYP706V12–V14, cytochrome P450 enzymes (CYPs) that exhibit a broader substrate range than SiCYP92B14. These enzymes oxidize (+)-sesamin and (+)-sesamolin during germination; when acting on (+)-sesamin, SiCYP706V12 produce (+)-sesaminol, whereas SiCYP706V13 and SiCYP706V14 yield (+)-episesaminone. The resulting oxidized lignans are then sequentially and regio-specifically glucosylated by UDP-glycosyltransferases (UGTs), including SiUGT73E4 and SiUGT73CH10 identified in this study, together with previously characterized UGTs. Functional and kinetic analyses revealed that these UGTs differentially process lignans with distinct molecular structures, thereby contributing to glycoside diversity. Notably, analysis of an SL-TG-deficient sesame line indicates that SiCYP706V12, rather than SiCYP92B14, plays a key role in SL-TG biosynthesis during seed development. Yeast two-hybrid assays revealed a physical interaction between SiCYP706V12 and a downstream UGT, suggesting a possible functional association between these enzymes in lignan metabolism. This underscores the overlapping yet distinct roles of CYP and UGT enzymes in coordinating lignan metabolism from seed development through germination. Our work highlights biochemical evolvability as a key factor in the specialization of plant metabolism in response to developmental and environmental cues.
A seismic reservoir permeability prediction approach based on gaussian process machine learning
Covalent inhibitors of human papillomavirus type 16 E6 protein restore p53 function and suppress growth of HPV-driven tumors in vivo
High-risk human papillomaviruses (HPVs) promote malignant progression through sustained expression of the viral oncoprotein E6, which drives degradation of the tumor suppressor p53 and creates an oncogenic dependency in HPV-positive cancers. Here, we identify a genotype-defined therapeutic vulnerability by selectively and irreversibly inactivating HPV-16 E6 through covalent targeting a cysteine proximal to its E6AP-binding interface. Pharmacologic inhibition of E6 restored p53 protein stability and transcriptional activity in HPV-16–positive cancer cells, inducing apoptosis and senescence while sparing HPV-negative epithelial cells. A CRISPR-engineered E6 cysteine-to-serine knock-in abolished compound activity in vitro and in vivo, establishing on-target mechanism. Transcriptomic profiling confirmed activation of p53-dependent tumor suppressor programs following E6 inactivation. In xenograft models of cervical and oropharyngeal cancer, irreversible inhibition of E6 suppressed growth of established tumors with minimal toxicity and no evidence of acquired resistance. These findings support covalent inactivation of HPV-16 E6 by a small molecule as a therapeutic strategy for HPV-associated malignancies.
Stage-dependent analysis of IL-6, TGF-β, osteocalcin, and sRANKL in the synovial fluid of dogs with osteoarthritis
Abstract Osteoarthritis (OA) is traditionally considered a degenerative joint disease. However, morphological similarities between OA progression and fracture healing suggest that OA may be a dysregulated bone repair process. This study investigated the morphological and biochemical parallels between OA and bone healing in dogs following cranial cruciate ligament rupture. To differentiate between biochemical profiles at different stages of the disease, a structured OA scoring system was developed based on radiographic and additional intraoperative findings. Synovial fluid samples from the OA cohort were analysed for IL-6, TGF-β, osteocalcin and sRANKL. IL-6 concentrations decreased significantly with advancing OA severity ( p = 0.0068), suggesting a shift from inflammatory to reparative mechanisms. TGF-β, osteocalcin and sRANKL demonstrated stage-dependent trends consistent with osteogenic and remodelling activity. However, these associations did not reach statistical significance. The combined morphological and biomarker findings suggest that OA progression exhibits features similar to the phases of bone healing, which are characterised by dynamic interactions between inflammation, resorption, and bone formation. Nevertheless, the predominantly non-significant biomarker correlations highlight the need for larger sample sizes and longitudinal studies to confirm these observations and further clarify causal mechanisms. These results support the concept that OA may represent a chronic but dysregulated adaptive repair response to joint damage, rather than a purely degenerative process.