143174 research outputs found
Sort by
Heterogeneous graph neural networks for assumption-based argumentation
No full textAssumption-Based Argumentation (ABA) is a powerful
structured argumentation formalism, but exact computation of extensions under stable semantics is intractable for large frameworks. We present the first Graph Neural Network (GNN) approach to approximate credulous acceptance in ABA. To leverage GNNs, we model ABA frameworks via a dependency graph representation encoding assumptions, claims and rules as nodes, with heterogeneous edge labels distinguishing support, derive and attack relations. We propose two GNN architectures—ABAGCN and ABAGAT—that stack residual heterogeneous convolution or attention layers, respectively, to learn node embeddings. Our models are trained on the ICCMA 2023 benchmark, augmented with synthetic ABAFs, with hyperparameters optimised via Bayesian search. Empirically, both ABAGCN and ABAGAT outperform a state-of-the-art GNN baseline that we adapt from the abstract argumentation literature, achieving a node-level F1 score of up to 0.71 on the ICCMA instances. Finally, we develop a sound polynomial time extension-reconstruction algorithm driven by our predictor: it reconstructs stable extensions with F1 above 0.85 on small ABAFs and maintains an F1 of about 0.58 on large frameworks. Our
work opens new avenues for scalable approximate reasoning in structured argumentation
Neighbourhood belonging, social cohesion, and mental wellbeing of children and parents in an ethnically diverse community sample in England
No full textBackground
Neighbourhood cohesion is considered an important and modifiable determinant of mental health that interacts with factors such as deprivation and ethnicity in complex ways. UK studies adequately representing ethnic minority groups are however scarce. We examined associations between neighbourhood belonging, social cohesion, and mental wellbeing of children and parents in an ethnically diverse community sample in England.
Methods
We analysed cross-sectional baseline data from the TOGETHER study, a randomised controlled trial testing the effectiveness of the ‘Strengthening Families, Strengthening Communities’ parenting programme developed to reach ethnic minority and other marginalised families living in England (ISRCTN: 15194500). Outcomes were parental mental wellbeing (Warwick-Edinburgh Mental Well-Being Scale, WEMWBS) and child socio-emotional difficulties (Strengths and Difficulties Questionnaire, SDQ). Neighbourhood belonging and social cohesion were assessed using the adapted Buckner scale. Multiple linear regression models were run, adjusted for sociodemographic factors including age, gender and ethnicity of the parent (for WEMWBS) or child (for SDQ); family socio-economic position; and family structure. Models assessing child socio-emotional difficulties additionally adjusted for parental mental wellbeing.
Results
The analysis sample included 638 participants with complete data, of whom 62% were from an ethnic minority background. Higher neighbourhood belonging and social cohesion were associated with higher parental mental wellbeing (higher WEMWBS scores) in fully adjusted models (β for neighbourhood belonging = 0.28, 95% CI: 0.17 to 0.40, p < 0.001; β for social cohesion = 0.49, 95% CI: 0.37 to 0.61, p < 0.001). Associations with WEMWBS were not moderated by ethnic group. Neighbourhood belonging was unrelated to child socio-emotional difficulties after adjustment for child and family characteristics. Higher social cohesion was associated with lower child socio-emotional difficulties after adjustment for covariates (β = -0.10, 95% CI: -0.20 to -0.01, p < 0.033), this association was fully attenuated after additional adjustment for parental mental wellbeing (β = 0.02, 95% CI: -0.07 to 0.12, p = 0.612).
Conclusions
In this diverse community sample, neighbourhood belonging and social cohesion were strongly related to parental mental health. Controlling for parental mental health explained the association between social cohesion and child socio-emotional difficulties. Fostering neighbourhood belonging and social cohesion may hold promise for efforts to improve both parent and child mental wellbeing
Heterologous COVID-19 vaccine schedule with protein-based prime (NVX-CoV2373) and mRNA boost (BNT162b2) induces strong humoral responses: results from COV-BOOST trial
No full textBackground
Heterologous schedules of booster vaccines for COVID-19 following initial doses of mRNA or adenoviral vector vaccines have been shown to be safe and immunogenic. There are few data on booster doses following initial doses of protein nanoparticle vaccines.
Methods
Participants of the phase 3 clinical trial of the COVID-19 vaccine NVX-CoV2373 (EudraCT 2020–004123-16) enroled between September 28 and November 28, 2020, who received 2 doses of NVX-CoV2373 administered 21 days apart were invited to receive a third dose booster vaccine of BNT162b2 (wild type mRNA vaccine) as a sub-study of the COV-BOOST clinical trial, and were followed up for assessment of safety, reactogenicity and immunogenicity to day 242 post-booster.
Results
The BNT162b2 booster following two doses of NVX-COV2373 was well-tolerated. Most adverse events were mild to moderate, with no serious vaccine-related adverse events reported. Immunogenicity analysis showed a significant increase in spike IgG titres and T-cell responses post-third dose booster. Specifically, IgG levels peaked at day 14 with a geometric mean concentration (GMC) of 216,255 ELISA laboratory units (ELU)/mL (95% CI 191,083–244,743). The geometric mean fold increase from baseline to day 28 post-boost was 168.6 (95% CI 117.5–241.8). Spike IgG titres were sustained above baseline levels at day 242 with a GMC of 58,686 ELU/mL (95% CI 48,954–74,652), with significant decay between days 28 and 84 (geometric mean ratio 0.58, 95% CI 0.53–0.63). T-cell responses also demonstrated enhancement post-booster, with a geometric mean fold increase of 5.1 (95% CI 2.9–9.0) at day 14 in fresh samples and 3.0 (95% CI 1.8–4.9) in frozen samples as measured by ELISpot. In an exploratory analysis, participants who received BNT162b2 after two doses of NVX-COV2373 exhibited higher anti-spike IgG at Day 28 than those who received homologous three doses of BNT162b2, with a GMR of 5.02 (95% CI: 3.17–7.94). This trend remained consistent across all time points, indicating a similar decay rate between the two schedules.
Conclusions
A BNT162b2 third dose booster dose in individuals primed with two doses of NVX-COV2373 is safe and induces strong and durable immunogenic responses, higher than seen in other comparable studies. These findings support the use and investigation of heterologous booster strategies and early investigation of heterologous vaccine technology schedules should be a priority in the development of vaccines against new pathogens
Investigating the potential of division of labour in synthetic bacterial communities for bioproduction
No full textWith recent advances in bioengineering, the ability to engineer organisms to perform desired functions has expanded significantly. However, adding new functions to a cell can increase metabolic burden and induce genetic instability, making engineered cells more prone to negative selection. This limitation of isogenic monocultures can be mitigated by designing synthetic microbial cocultures with division of labour, in which metabolic pathways are distributed across multiple organisms to reduce the load on any single cell. Such an approach can lead to higher product titres by freeing more cellular resources for the target pathway and enables the biosynthesis of complex compounds requiring multi-step pathways that are difficult to achieve in monocultures. Maintaining stable cocultures over long periods requires a control system to prevent one population from dominating the other.
In this thesis, the potential of division of labour for the bioproduction of high-value compounds using synthetic bacterial communities was explored. Two coculture systems were constructed: an Escherichia coli coculture producing violacein and a Pseudomonas putida coculture producing limonene and perillyl alcohol, with subpopulations expressing different segments of their respective metabolic pathways. For violacein production, several pathway-splitting strategies were tested by partitioning the pathway at different intermediates, and culture conditions were optimised to maximise production. Coculture performance exceeded that of monocultures when three conditions were met: efficient exchange of intermediates, similar growth rates between subpopulations, and balanced initial inoculation ratios. For P. putida, two strains were engineered—one producing limonene and the other converting it to perillyl alcohol. These were optimised for D- and L-enantiomers of both compounds, and cocultures consistently outperformed monocultures. Finally, a control system was proposed for two-species microbial communities, decoupling growth and ratiometric control. Two quorum-sensing systems were characterised for communication, and three bacteriocins were evaluated for growth regulation, culminating in an integrated control circuit design.Open Acces
Event-based simulation of stochastic memristive devices for neuromorphic computing
No full textIn this paper, we build a general modelling framework for memristors, suitable for the simulation of event-based systems such as hardware spiking neural networks, and more generally, neuromorphic computing systems composed of three independent components: i) an event-based modelling approach, extending and generalising an existing general model of memristors - the Generalised Metastable Switch Model (GMSM) [1] - eliminating errors associated with discrete time approximation, as well as offering potential improvements in terms of suitability for neuromorphic memristive system simulations; ii) a volatility state variable to allow for the unified understanding of disparate non-linear and volatile phenomena, including state relaxation, structural disruption, Joule heating, and non-linear drift in different memristive devices; and iii) a readout equation that separates the latent state variable evolution from explicit variables of interest such as an instantaneous resistance. We exhibit an illustrative implementation of this framework, fit to a resistive drift dataset for titanium dioxide memristors, based on a proposed linear conductance model for resistive drift in the devices. Finally, we highlight the application of the model to neuromorphic computing, through demonstrating the contribution of the volatility state variable to switching dynamics, resulting in frequency-dependent switching (for stable memristors acting as programmable synaptic weights) and the generation of action potentials (for unstable memristors, acting as spike-generators)
Disentangling neural disjunctive normal form models
No full textNeural Disjunctive Normal Form (DNF) based models are powerful and interpretable approaches to neuro-symbolic learning and have shown promising results in classification and reinforcement learning settings without prior knowledge of the tasks. However, their performance is degraded by the thresholding of the post-training symbolic translation process. We show here that part of the performance degradation during translation is due to its failure to disentangle the learned knowledge represented in the form of the networks’ weights. We address this issue by proposing a new disentanglement method; by splitting nodes that encode nested rules into smaller independent nodes, we are able to better preserve the models’ performance. Through experiments on binary, multiclass, and multilabel classification tasks (including those requiring predicate invention), we demonstrate that our disentanglement method provides compact and interpretable logical representations for the neural DNF-based models, with performance closer to that of their pre-translation counterparts. Our code is available at https: //github.com/kittykg/disentangling-ndnf-classification
A subset of haematopoietic stem cells resists Plasmodium infection-induced stress by uncoupling interferon sensing and metabolic activation
No full textPreprint versio
Potential health and cost impacts of a point-of-care test for neonatal sepsis and possible serious bacterial infections in infants: a modeling analysis in two settings
No full textBackground
Sepsis contributes to nearly 50% of neonatal deaths in resource-limited settings. Accurate, timely diagnosis can improve outcomes, reduce inappropriate antibiotic use, and save healthcare costs. We aimed to determine the minimum technical requirements and cost of a point-of-care test (POCT) for neonatal sepsis to be clinically effective in hospitals and community settings in low-resource settings.
Methods
We modeled the diagnosis and treatment of hospitalized neonates and infants at primary care facilities with suspected sepsis. Health outcomes (mortality, hospital stays, and healthcare-associated infections [HAIs]) were compared under empiric treatment with varied blood culture scenarios to a POCT. Test performance, bacterial infection prevalence, and discharge criteria were varied. A threshold economic analysis identified maximum allowable costs for a cost-neutral POCT.
Results
A POCT could reduce neonatal deaths significantly at hospitals (up to 19%) and community levels (up to 70%), compared with baseline, by enabling faster therapy initiation and reducing unnecessary hospitalizations and HAIs. Healthcare costs could drop by 17%–43% in hospitals and 48%–81% in primary care settings. A POCT priced at 3 for community use could remain cost-neutral.
Conclusions
A POCT for neonatal sepsis, even with moderate accuracy, could improve outcomes by accelerating diagnosis, supporting antibiotic stewardship, and lowering costs. High sensitivity is essential to minimize deaths from missed diagnoses and delayed antibiotic therapy. Our results suggest 85% sensitivity and 80% specificity for hospitalized neonates, and 90% sensitivity and 70% specificity in primary care settings are the minimum necessary technical requirements for the POCT
General practice antibiotic prescriptions attributable to respiratory syncytial virus by age and antibiotic class: an ecological analysis of the English population
No full textBackground
Respiratory syncytial virus (RSV) may contribute to a substantial volume of antibiotic prescriptions in primary care. However, data on the type of antibiotics prescribed for such infections are only available for children <5 years in the UK. Understanding the contribution of RSV to antibiotic prescribing would facilitate predicting the impact of RSV preventative measures on antibiotic use and resistance. The objective of this study was to estimate the proportion of antibiotic prescriptions in English general practice attributable to RSV by age and antibiotic class.
Methods
Generalized additive models examined associations between weekly counts of general practice antibiotic prescriptions and laboratory-confirmed respiratory infections from 2015 to 2018, adjusting for temperature, practice holidays and remaining seasonal confounders. We used general practice records from the Clinical Practice Research Datalink and microbiology tests for RSV, influenza, rhinovirus, adenovirus, parainfluenza, human metapneumovirus, Mycoplasma pneumoniae and Streptococcus pneumoniae from England’s Second Generation Surveillance System.
Results
An estimated 2.1% of antibiotics were attributable to RSV, equating to an average of 640 000 prescriptions annually. Of these, adults ≥75 years contributed to the greatest volume, with an annual average of 149 078 (95% credible interval: 93 733–206 045). Infants 6–23 months had the highest average annual rate at 6580 prescriptions per 100 000 individuals (95% credible interval: 4522–8651). Most RSV-attributable antibiotic prescriptions were penicillins, macrolides or tetracyclines. Adults ≥65 years had a wider range of antibiotic classes associated with RSV compared with younger age groups.
Conclusions
Interventions to reduce the burden of RSV, particularly in older adults, could complement current strategies to reduce antibiotic use in England
Clustering lung function and symptom profiles for asthma risk stratification
No full textAsthma is a heterogeneous condition often studied through wheeze alone, yet the interplay between lung function and reported symptoms remains underexplored. To capture this heterogeneity, we applied Bayesian Profile Regression to data from school-age children in two prospective birth cohorts, integrating airway hyperresponsiveness, lung function, bronchodilator reversibility, allergic sensitisation, reported symptoms, and physician diagnosis.
In the Manchester Allergy and Asthma Study (discovery cohort), five reproducible clusters were identified: HA-LLF (high asthma-low lung function), HA-NLF (high asthma-near-normal lung function), LA-LLF (low asthma-low lung function), LA-NLF (low asthma-normal lung function), and INT-SYM (intermediate asthma with prominent symptoms). The HA-LLF and HA-NLF clusters had very high asthma prevalence (80–100%), but differed markedly in lung function, airway responsiveness, bronchodilator reversibility, sensitisation, and symptom burden. The LA-HLF and LA-NLF clusters with low asthma prevalence (<5%) displayed contrasting lung function profiles, while INT-SYM (~50% asthma prevalence) was largely defined by prominent symptoms such as chest tightness and shortness of breath. These subtypes were replicated in an independent cohort, Isle of Wight. Our findings demonstrate that integrating physiological, immunological, and symptom-based measures yields clinically meaningful asthma subtypes beyond wheeze-based definitions and may support more precise disease classification