143174 research outputs found
Sort by
Risks of placing complex implants in a general healthcare system- lessons from real-life experience
No full textIntroduction: Surgical implants, such as joint replacements, are used for many serious conditions. Innovation continues to supply new implants, including outputs of the soft robotics revolution. However, they carry risk of complications with potentially devastating consequences.
Methods: We present an opinion paper providing the reflections of two surgical technologists on present challenges to safety, efficacy and broad implementation of medical implants.
Results: We highlight lack of familiarity with implant surgery in healthcare services, with concomitant risk. First-in-human application of new implants is not sufficiently standardised and regulated. IDEAL-D is a structured framework for medical devices (Idea, Development, Exploration, Assessment, Long-term study). Once CE-marked and approved for mainstream use, there are problems with the implementation. “Early adopter” surgeons and centres face cultural inertia, lack of funding support and issues around training, especially learning curves. Patient selection may not be well-defined, and complications inaccurately reported, affecting implant dissemination detrimentally. The Cumberlege report showed how harmful this can be.
Conclusion: There is need to standardise early clinical studies. Implementation of implantable devices requires changes to whole-team training, funding and post-implementation reporting. The IDEAL-D framework represents an important step, but other system-wide changes are required if implants are to achieve their intended clinical impact
Optimizing scheduling in dual-pulse nucleoside labeling experiments for cell cycle analysis
No full textAll eukaryotic cells go though a universal sequence of phases during their division cycle, where the phase
timings vary according to cell type and state. Dual-pulse nucleoside labeling (DPNL) is a standard, widely applicable
experimental DNA base substituting technique to probe cell cycle kinetics at the population level, including in living organisms. In such an experimental protocol, a key scheduling parameter is the choice of waiting time between the two labeling pulses. Here, we model population cell cycle dynamics as a three-stage Poisson process with an idealized S-phase labeling step, and use a simulation-based look-up procedure to demonstrate that the inter-pulse waiting time can be optimized to maximize the signal-to-noise ratio of inferred cycle parameters — an issue that is especially critical in DPNL experiments with limited cell numbers and replicates. An optimal choice of pulse scheduling typically improves S phase time inference by 50% compared to a random choice. We further discuss the procedure to perform such a task in an experimentally relevant
setting
Sub-picosecond permittivity of carbon nitrides probed with terahertz spectroscopy: revealing high dielectric response and conductivity
No full textOrganic based semiconductor materials offer emerging and sustainable solutions for solar energy conversion technologies and electronics. However, knowledge of their intrinsic (photo)physical properties is often limited, especially the effect of dielectric properties (εr = ε′) on exciton separation and charge generation at sub-picosecond timescales, which corresponds to THz frequencies. Thus, THz Time Domain Spectroscopy (THz-TDS) is used to directly and accurately extract the complex permittivity (ε = ε′ + iε″) and THz conductivity (σTHz) of organic Carbon Nitrides (CNx) and other polymers and elucidate the influence of environmental humidities. Overall, the THz dielectric response ε′ of CNx surpasses other organic and even glycolated materials, and water. For the ionic and 2D carbon nitride K-PHI, complex permittivity ε was observed to be strongly humidity dependent, with both ε′ and σTHz doubling from dry to humid conditions (ε′ from ∼4 to 8, σTHz 75 to 150 S/m, respectively). When compared to other photocatalysts, the THz dielectric response of CNx and especially humid K-PHI is within range of well-known oxides such as TiO2 that can efficiently generate charges from excitons, due to low exciton binding energies resulting from high ε′. The importance of dielectric property characterization on functionally relevant frequencies is thus highlighted, especially in the THz gap (0.1 – 10 THz), to understand the photophysical behaviour of organic semiconductors, even in the presence of water and hydrated ions. Such THz complex permittivity determination may also be beneficial for exploring next generation photo(electro)catalysts, electronics or ionotronics, and for computational property predictions that often require knowledge of ultrafast photophysical properties
Synthesizing epileptic seizures: Gaussian processes for EEG generation
No full textPreprint versionReliable seizure detection from electroencephalography (EEG) time series is a highpriority clinical goal, yet the acquisition cost and scarcity of labeled EEG data limit the performance of machine learning methods. This challenge is exacerbated by the long-range, highdimensional, and non-stationary nature of epileptic EEG recordings, which makes realistic data generation particularly difficult. In this work, we revisit Gaussian processes as a principled and interpretable foundation for modeling EEG dynamics, and propose a novel hierarchical framework, GP-EEG, for generating synthetic epileptic EEG recordings. At its core, our approach decomposes EEG signals into temporal segments modeled via Gaussian process regression, and integrates a domain-adaptation variational autoencoder. We validate the proposed method on two real-world, open-source epileptic EEG datasets. The synthetic EEG recordings generated by our model match real-world epileptic EEG both quantitatively and qualitatively, and can be used to augment training sets
Representation learning for efficient reinforcement learning
No full textReinforcement-learning (RL) agents learn from often unstructured, incomplete or complex observations collected from their environment.
In traditional RL approaches, the burden of interpreting these observations falls on the value-function and policy models, which use rewards as the only learning signal.
Simultaneous representation learning, i.e., interpreting observations, and RL, therefore, becomes a challenging task in complex or partially observable environments.
This thesis improves upon traditional approaches by proposing a separation of representation learning and RL.
We cast these as separate tasks and propose learning representations using more meaningful signals than rewards alone, such as observed states, actions and environment dynamics.
These representations are then used in RL as a downstream task.
First, we focus on the single-agent domain.
RL algorithms in this context often struggle to learn strong policies in complex environments with many states and actions.
To alleviate this problem, we propose an embedding methodology, which jointly learns representations for states and actions.
These representations capture structure in the environment and can be used to compress the original state and action spaces, reducing the complexity of the environment.
We use a model of the environment and state information instead of reward as a learning signal for representation learning.
These representations are then used in policy-gradient-based RL algorithms, thereby combining elements of model-free and model-based RL.
We demonstrate the validity of the approach theoretically and empirically.
Second, we turn to the multi-agent domain, where we also propose a separation of representation learning and RL.
However, here we use representation learning to infer information missing from agents' local observations.
We first train a belief model using complete environment information, which is then used by a fully decentralised state-based RL algorithm using local observations only.
We construct partially observable environments and empirically demonstrate our approach's efficacy compared to relevant benchmarks.Open Acces
PLAT-M8 prognostic biomarker, its related genes, and mechanistic insights in ovarian cancer: A systematic review with bioinformatic validation
No full textOvarian cancer is a leading cause of gynecologic cancer-related death, necessitating reliable prognostic
biomarkers. PLAT-M8 has emerged as a promising DNA-methylation biomarker influencing key genetic
pathways in relapsed ovarian cancer; however, its associated gene expression and the downstream
mechanisms of PLAT-M8 methylation in primary ovarian cancer (particularly in driving chemoresistance and
recurrence) remain less well understood. A systematic review of preclinical and clinical studies (2008–2023)
was conducted to explore the prognostic potential of genes associated with PLAT-M8 loci in ovarian cancer,
resulting in 27 studies for inclusion. Bioinformatics tools (KM-plotter, TNMplot, Cancer Hallmark) were used to
validate the clinical relevance of PLAT-M8 genes [cg05529343 (ZNF385D), cg12992827 (ZPLD1), cg16172923 (MAD1L1), cg07960624 (SAMD12), cg25953130 (ARID5B), cg13691961 (DUSP6), cg01692018 (PPP2R5E), and cg07573872 (SBNO2)] by analysing overall survival (OS), progression-free survival (PFS), gene
correlations, and tumour vs. normal tissue. As signature, PLAT-M8-associated genes were significantly
associated with OS (HR 1.38, 95%CI: 1.13-1.70, p=0.0016) and PFS (HR 1.38, 95%CI: 1.14-1.66, p=0.0008).
These findings were consistent across subgroup analyses accounting for treatment, surgery, stage, and
histological type. Notably, strong correlations between PLAT-M8 methylation and gene expression were
observed, with DUSP6, ZNF385D, and ARID5B emerging as key prognostic markers. These results highlight the prognostic value of PLAT-M8 to inform personalised treatment strategies and identify new therapeutic targets, aiming to improve outcomes for ovarian cancer patients
Changes in admissions, care processes and outcomes for very and extremely preterm infants in England and Wales: an 11-year whole population study
No full textIntroduction: Preterm birth is a major public health concern with lifelong consequences. National policies in England and Wales aim to reduce deaths, improve outcomes, and address disparities. We examined changes in admissions, care processes, and outcomes among extremely (EPT) and very preterm (VPT) infants overall and by maternal ethnicity.
Methods: We conducted a retrospective, population-based cohort study using the National Neonatal Research Database (NNRD) and Office for National Statistics (ONS) data, covering all National Health Service neonatal units from 2013–2023. Temporal trends in care processes and clinical outcomes were estimated using modified Poisson regression. Models were adjusted for gestational age, sex, multiplicity, and birthweight z-score. Results are presented as adjusted risk ratios (aRR) with 95% confidence intervals (CI).
Results: There were 26,132 EPT and 55,789 VPT admissions, with no significant change in overall admission rates. Admissions of babies born below 24 weeks gestation almost doubled. Mothers of Black ethnicity consistently had the highest rates of EPT and VPT admissions. Deliveries by emergency Caesarean section increased (EPT: 1.04, 1.03–1.04; VPT: 1.02, 1.01–1.02). Delivery room intubation (EPT: 0.96, 0.96–0.97; VPT: 0.93, 0.93–0.93), intubated respiratory support (EPT: 0.99, 0.99–0.99; VPT: 0.96, 0.96–0.97), and surgical or device closure of patent ductus arteriosus (PDA) closure (EPT: 0.82, 0.80–0.84; VPT: 0.86, 0.81–0.92) decreased. In EPT infants, maternal milk uses at discharge (1.01, 1.01–1.02) and late-onset bloodstream infection increased (1.03, 1.02–1.04) and early postnatal transfers (0.98, 0.97–0.99) and mortality decreased (0.98, 0.97–0.99). In VPT infants severe necrotising enterocolitis (NEC) (0.96, 0.94–0.98) and survival without major morbidity decreased (0.99, 0.99–0.99).
Conclusions: Over the last decade, EPT and VPT admission rates in England and Wales have not changed and ethnic disparities remain evident. Delivery of some evidence-based care processes improved, but clinical outcomes showed only modest gains
ITRF/LhARA conceptual design report
No full textThis document, the principal deliverable of the Preliminary Activity, presents the conceptual design for LhARA to serve the ITRF. The conceptual design of the accelerator facility is presented in Chapter 2. Comparison of the laser-hybrid solution with the conventional alternatives has allowed the LhARA approach to be confirmed as the baseline design [15]. The conceptual design for the facility is shown in figure 1
Differentiable rendering for dense visual SLAM
No full textThis thesis investigates differentiable rendering for dense visual Simultaneous Localization and Mapping (SLAM), specifically examining how novel rendering algorithms and 3D scene representations enhance reconstruction and spatial understanding. The choice of scene representation is fundamental to the efficiency of 3D modeling. Traditional approaches, which rely on direct 3D measurements from sensors or predictions from pre-trained networks, often struggle with limited operational ranges or training data biases. This necessitates 3D representations that can be flexibly optimized using real-time 2D observations.
With advancements in computing hardware, 3D representations optimized through differentiable rendering have gained prominence. These methods achieve high-fidelity inverse rendering and novel view synthesis. While early techniques required lengthy optimization, the introduction of explicit data structures has shifted performance toward real-time, enabling integration into SLAM. Unlike offline reconstruction, online SLAM requires continuous estimation and updating of unknown environments from sequential data, posing unique challenges for optimization stability and memory management.
This thesis examines the use of 3D scene representations in combination with differentiable rendering pipelines to enable real-time SLAM systems. The work begins by developing classical depth map based SLAM methods to establish a foundation in existing standard pipelines. Building on this baseline, the thesis explores advanced scene representations — specifically NeRF and the latest 3D Gaussian Splatting techniques — as novel approaches to SLAM. Through detailed experiments and analysis, the thesis demonstrates how to effectively address the open-world challenges inherent in SLAM tasks using these modern representations. It also highlights the unique advantages that differentiable rendering offers, such as improved accuracy and flexibility, positioning it as a promising direction for the future of visual SLAM.Open Acces
Synergies between energy system decarbonisation and air quality in the UK
No full textEnergy system decarbonisation strategies are expected to improve UK air quality, yet these two areas have largely been studied in isolation. The limited existing research base has prevented a holistic understanding of the air pollution impacts of future energy strategies. This thesis uses modelling tools such as the UK Integrated Assessment Model to better understand these air pollution and energy system decarbonisation synergies.
The first half of the thesis examines potential policy implications. A screening study assesses emissions and contributions to population-weighted mean concentrations (PWMC), highlighting how decarbonisation pathways influence future air pollution abatement measures. A second study focuses on black carbon, a pollutant with both health and climate impacts, comparing emissions from two inventories based on different methodologies. Biomass combustion emerges as a key concern for future black carbon emissions.
The second half of the thesis starts with a detailed modelling assessment of uncertainty and sensitivity in future scenarios. Results did not show a significant improvement in PM2.5 PWMC contributions from the energy system by 2050 when comparing net zero-only futures to broader energy scenarios with sustained fossil fuel usage. Sensitivity analysis using Sobol indices identifies biomass combustion, hydrogen production, and road transport tyre wear as dominant sources of uncertainty in these PM2.5 projections.
Finally, spatial modelling of UK future energy scenarios reveals similar PM2.5 concentrations across scenarios with varying levels of hydrogen deployment, with most areas falling below 6 µg m-3. Total PM2.5 PWMC drops from 8.73 µg m-3 in 2020 to between 5.98 and 6.41 µg m-3 in 2050. The largest potential hydrogen economy contributes 0.3 µg m-3 to PM2.5 PWMC, indicating minimal risk to air quality targets.
The findings in this thesis underline the importance of integrated modelling and collaboration between air quality and climate communities to achieve the greatest co-benefits for health and the environment.Open Acces