53417 research outputs found
Sort by
PROthrombin complex concentrate versus fresh frozen Plasma for bleeding in adults undergoing HEart SurgerY (PROPHESY-2 trial): a phase III, randomised control trial in England and Wales
Get PDFInterleukin-36 upregulates type-I interferon responses in systemic lupus erythematosus by promoting the accumulation of self-nucleic acids
Get PDFIntroduction: Several studies have reported an up-regulation of interleukin (IL)-36 in the serum of patients with systemic lupus erythematosus (SLE). Here, we sought to define the mechanisms whereby IL-36 may contribute to the over-activation of type I Interferon (IFN) responses observed in SLE.
Methods: We carried out single-cell (sc)RNA-seq in healthy peripheral blood mononuclear cells treated with IL-36 (n=5 donors). We compared the genes and transcriptional networks that were induced by IL-36 with those that were upregulated in a published SLE scRNA-seq dataset (n=33 cases and 11 controls). In follow-up studies, we validated the effects of IL-36 on monocytes by real-time PCR (n=9 donors) and flow-cytometry (n=6).
Results: Classical monocytes were the immune population most affected by IL-36 treatment (n=203 Differentially Expressed Genes). In these cells, IL-36 upregulated transcriptional networks (regulons) driven by IRF7, a key activator of type I IFN responses. A similar upregulation of IRF7 regulons was observed in the monocytes of SLE cases, where measurements of IL-36 and IRF7 activity were significantly correlated (r=0.35, P = 0.02). Experimental follow-up studies in human monocytes showed that IL-36 downregulates multiple RNAse genes (RNASE1, RNASE6, RNASET2). IL-36 treatment of monocytes also increased the percentage of apoptotic cells (45% vs 37% in untreated cells; P = 0.001), which are a critical source of self-nucleic acids.
Conclusion: We find that IL-36 promotes monocyte apoptosis while downregulating self-nucleic acid clearance. Thus, IL-36 contributes to the accumulation of self-nucleic acids, a key driver of type I IFN responses in SLE
Artificial intelligence tools for automating evidence synthesis: A scoping review (Preprint)
No full textChromosomal Instability as a Driver of cGAS-STING Dysfunction in High Grade Serous Ovarian Cancer
Get PDFChromosomal instability (CIN) is defined as the continual gain or loss of chromosome fragments or whole chromosomes and is a feature of high grade serous ovarian cancer (HGSOC). Numerous drivers of CIN in HGSOC have been identified including homologous recombination repair deficiency (HRD). One consequence of ongoing CIN is the accumulation of cytoplasmic self-DNA which promotes the activation of DNA sensing pathways such as cGAS-STING. Acute cGAS-STING activation by DNA damaging agents has been associated with immune activation and synergy with immunotherapy. Recently, several publications have demonstrated a pro-tumorigenic role for chronic cGAS-STING signalling in CINhigh models of various cancers. The consequences of chronic cGAS-STING activation in homologous recombination repair deficient (HRD) HGSOC remains unexplored. Therefore, we investigated the differences in cGAS-STING functionality in CINlow and CINhigh HGSOC-representative murine cell lines, in addition to evaluating the prevalence of cGAS-STING dysfunction in panels of human HGSOC cell lines. CINhigh HSGOC murine cells demonstrated reduced basal STING mRNA and protein expression in comparison to CINlow cells. As a consequence, CINhigh cells failed to induce PD-L1 and cytokines in response to STING agonism. This phenotype was mimicked in CINlow cell lines through repetitive STING agonism. We hypothesised that STING downregulation would be reversed upon alleviation of basal cGAS activation in CINhigh cells and indeed, evidence of this was observed upon cGAS knockout. However, this phenotype was not consistent across knockout clones, which suggests heterogeneity within the parental population with regard to how STING is being downregulated. Previously published CIN-induced cGAS-STING phenotypes, including metastasis-related and NFκB-related signatures, were not recapitulated in the CINhigh murine model. In human HGSOC cell lines, extensive cGAS-STING dysfunction was observed irrespective of CIN rate, which suggests that CIN is not the sole contributing factor to cGAS-STING dysfunction in HGSOC cell line models. Overall, it was determined that cGAS-STING dysfunction is apparent in CINhigh mouse and human cell line models of HGSOC, though important differences from previous studies using different mechanistic drivers of CIN were noted. Future work aims to explore the mechanisms mediating STING suppression in the CINhigh models, and to produce further CINhigh models to evaluate if cGAS-STING dysfunction becomes apparent over time as CIN increases
TinyFed6G: Federated Learning With TinyML for Resource-Constrained Intelligence in 6G Edge Networks
Get PDFThe rising prevalence of ultra-low-power microcontrollers in intelligent edge devices has created a demand for collaborative learning on devices. Federated Learning (FL) platforms are designed for edge devices; however, their functionality may be limited due to edge device constraints on memory, energy, and communication bandwidth, particularly in non-IID (Non-Independent and Identically Distributed) data distributions and heterogeneous hardware. To overcome these issues, this paper proposes TinyFed6G, a communication-efficient hierarchical FL framework that is 6G-aligned with TinyML (Tiny Machine Learning) clients. The architecture supports a dual-mode execution that enables the device to dynamically switch between training mode and inference mode based on real-time profiling of the device. Enhanced communication efficiency is achieved through semantic-aware compression, which filters quantized model updates based on a cosine similarity threshold. Personalized learning is accomplished by selectively fine-tuning the head layer of the model. The performance of the framework is evaluated using distributions. Experimental results demonstrate that TinyFed6G achieves a final accuracy of 86.2%, a 6.5% personalization gain
Unsteady Vortex Characterisation for Supersonic Flow Experiments
No full textThe characterisation of vortices in supersonic flow experiments is often carried out with time-averaged measurements, however this can result in errors due to the inherent unsteadiness of the vortex. The meandering of the vortex causes its position and angle to the mean flow to vary over time, which ‘smears’ the velocity profiles obtained using time-averaged measurement techniques such that the measured peak tangential velocity and core radius differ from their true values. This paper presents a methodology for characterising the meander of vortices in supersonic flows, allowing for the correction of errors in time-averaged measurements due to unsteadiness in the vortex position and angle to the mean flow. These approaches are demonstrated for a series of vortices generated in a Mach 1.5 freestream, with swirl ratios in the range 0.1 - 0.2 and vortex Reynolds numbers based on core diameter in the range 70,000 - 320,000. Time-averaged tangential velocity profiles are measured with traversed laser Doppler velocimetry and high-speed schlieren image sequences are analysed to obtain time-resolved data for the unsteady vortex position and angle. The resulting statistics are used to correct the measured tangential velocity profiles, identifying errors of up to 68% in the peak velocities and 50% in the core radii. The analysis shows that errors due to variation of the vortex centre position are in general significantly greater than those due to variation of the angle to the mean flow
Machine learning-enabled uncertainty quantification for thermo-catalytic reactors: A study on fugitive methane oxidation in monolith reactors
No full textUltra-lean methane oxidation via catalytic combustion is critical for mitigating greenhouse gas emissions from fugitive methane sources. However, the catalytic oxidation process exhibits significant uncertainties that hinder its widespread implementation. To address this challenge, the present study develops a robust machine learning-based framework for quantifying combustion uncertainties, enabling more effective emission control strategies. The work presents a novel hybrid methodology integrating polynomial chaos expansion (PCE) with artificial neural networks (ANN), achieving real-time prediction of methane conversion rates and their uncertainties in monolith reactors. The machine learning model reduces computational time from hours to seconds while achieving excellent agreement with detailed 1D plug-flow reactor simulations. The investigation reveals that variations in methane concentration (0.2 %–1.3 %,
10 %), inlet temperature (800–1000 K,
2 %), and inlet velocity (0.8–1.2 m/s,
5 %) significantly influence conversion uncertainty, with inlet temperature identified as the dominant parameter (CV
75 %). Stability improves at elevated temperatures (
950 K) and lower flow velocities (CV
10 %) compared to higher velocities (CV = 17 %–22 %). Additionally, catalyst deactivation, represented by reduced coating length, decreases methane conversion rates and increases uncertainty, with longer coatings providing greater stability at higher inlet temperatures. This work advances the fundamental understanding of uncertainty propagation in ultra-lean catalytic methane combustion and establishes a generalisable, computationally efficient PCE-ANN framework applicable to catalytic combustion of diverse fuels
Neuromodulation of a peripheral nerve using fully polymeric cuff electrodes: Understanding predictability of selective stimulation.
No full textObjective
Peripheral nerve stimulation (PNS) offers therapeutic benefits across numerous clinical applications but remains limited by poor spatial selectivity in mixed nerves. This study aimed to evaluate whether a fully polymeric, transverse, multipolar nerve cuff can achieve selective fascicular activation and to assess the predictability of such selectivity using imaging-informed computational models.
Approach
A flexible, fully polymeric nerve cuff fabricated from a conductive elastomer was developed and evaluated ex vivo on rat sciatic nerves. Compound nerve action potentials were recorded from individual fascicles to quantify selectivity across a wide range of stimulation parameters. The non-metallic electrodes enabled microCT-based three-dimensional reconstruction of nerve-electrode geometries without imaging artefacts, which were incorporated into anatomically accurate simulations using the ASCENT modeling pipeline.
Main results
Ex vivo experiments demonstrated reliable neural recordings and high levels of fascicular selectivity (selectivity index > 0.65 in each fascicle). Imaging-informed simulations reproduced selective activation patterns in some cases but showed systematic discrepancies in both selectivity magnitude and electrode-fascicle correspondence, particularly for sural and tibial fascicles. Simulated outcomes were more sensitive to neuroanatomical variability than experimental results, highlighting limitations in current modeling assumptions.
Significance
These findings validate fully polymeric conductive elastomer cuffs as effective alternatives to metallic nerve interfaces for selective peripheral nerve stimulation. The study also demonstrates the value of combining microCT imaging with computational modeling to interrogate and refine predictive frameworks, underscoring the need for improved tissue and electrode modeling to advance spatially selective PNS technologies.

Advancing Embodied Intelligence in Robotic-Assisted Endovascular Procedures: A Systematic Review of AI Solutions.
No full textEndovascular procedures have revolutionized vascular disease treatment, yet their manual execution is challenged by the demands for high precision, operator fatigue, and radiation exposure. Robotic systems have emerged as transformative solutions to mitigate these inherent limitations. A crucial moment has arrived, where a confluence of pressing clinical needs and breakthroughs in AI creates an opportunity for a paradigm shift toward Embodied Intelligence (EI), enabling robots to navigate complex vascular networks and adapt to dynamic physiological conditions. Data-driven approaches, leveraging advanced computer vision, medical image analysis, and machine learning, drive this evolution by enabling real-time vessel segmentation, device tracking, and anatomical landmark detection. Reinforcement learning and imitation learning further improve navigation strategies and replicate expert techniques. This review systematically analyzes the integration of EI into endovascular robotics, identifying challenges such as the heterogeneity in validation standards and the gap between human mimicry and machine-native capabilities. Based on this analysis, a conceptual roadmap is proposed that reframes the ultimate objective away from systems that supplant clinical decision-making. This vision of augmented intelligence, where the clinician's role evolves into that of a high-level supervisor, provides a principled foundation for the future of the field