27716 research outputs found
Sort by
Multimorbidity and myocardial infarction: an investigation using linked, routinely collected health record data
No full textMultimorbidity, defined as the co-occurrence of two or more long-term conditions (LTCs), is recognised as a key factor influencing patient management and prognosis in a number of contexts. This thesis sought to examine patterns of pre-existing multimorbidity in individuals with acute myocardial infarction (MI) and evaluate their impact on the receipt of guideline-directed care and long-term clinical outcomes.
Firstly, a systematic review and meta-analysis summarised the existing literature, highlighting the need for comprehensive ascertainment of LTCs. Subsequently, an algorithm was developed to ascertain 321 LTCs (aggregated into 116 clinical phenotypes), guided by multidisciplinary clinical, patient and public input – the Inclusive Multimorbidity Phenotyping Algorithm and Codelist Tool (IMPACT). A nationally representative, retrospective cohort study of individuals presenting with an index MI was conducted using linked primary and secondary care data. Latent class analysis (LCA) identified clusters of co-occurring LTCs (‘multimorbidity endotypes’). Flexible parametric survival models estimated 10-year cause-specific cumulative incidence ratios (CIRs) and corresponding 95% confidence intervals (CI) for each endotype, adjusting for covariates and accounting for competing risks.
Among 272,344 patients with MI (171,042 with non-ST-elevation MI [NSTEMI] and 101,302 with ST-elevation MI [STEMI]), 91.1% had ≥ 2 pre-existing LTCs (median: 7; interquartile range: 4 to 12). A greater number of LTCs were observed with female sex, increasing age, socioeconomic deprivation and NSTEMI. The number of pre-existing LTCs at MI presentation increased substantially over the study period. There was significant heterogeneity in receipt of guideline-directed care by multimorbidity endotype. The most adverse endotype (‘diabetes-related multimorbidity’) was associated with greatest 10-year mortality (NSTEMI: CIR 1.54, 95% CI: 1.48 – 1.60, and STEMI: CIR 1.60, 95% CI: 1.49 – 1.71), compared with the most favourable endotype (‘background low-impact disease’). In older adults with NSTEMI, benefits of invasive treatment strategy on clinical outcomes were observed among some endotypes (‘background low-impact disease’, ‘mental and behavioural health-related multimorbidity’), but not others.
This study demonstrated that individuals with MI had a high burden of pre-existing multimorbidity, which determined their clinical management and outcomes. This underscores the need to consider the full spectrum of pre-existing long-term conditions: a ‘whole patient’ approach to the management of MI is required
Rate Splitting in Laser-based Optical Wireless Networks
No full textOptical wireless communication (OWC) systems using infrared lasers as transmitters, offer high capacities compared to radio frequency (RF) networks. In this work, the primary focus is on interference management in such systems.
The work introduces for the first time rate splitting (RS) in a laser-based OWC system. Multiple optical access points (APs) in a laser-based system can serve multiple users simultaneously by splitting the message of a user into common and private messages, each message with a certain level of power, while on the other side users decode their messages following a specific methodology. Interestingly, the power must be carefully allocated between these messages to minimize multi-user interference (MUI) and maximize the spectral efficiency of the network. The RS strategy offers higher sum rates compared to traditional interference management schemes. However, in scenarios with high number of users, the performance of RS faces severe limitations due to noise enhancement.
To address this challenge, a two-tier precoding RS scheme called hierarchical rate splitting (HRS) is proposed for a multi-user signal cell OWC network to enhance the sum rate and alleviate channel state information (CSI) requirements. In HRS, users are divided into multiple groups, and outer RS is applied to manage inter-group interference, while inner RS manages intra-group interference. This methodology requires a new message, referred to as the outer common message, to manage inter-group interference. Therefore, the power budget must be used efficiently among the three messages to maximize the sum data rate of the network. In this context, an optimization problem is formulated for power allocation under certain constraints, and the results demonstrate that HRS achieves high performance in dense OWC networks compared to RS and OMA.
In laser-based OWC networks, users might experience severe ICI due to the confined coverage area of the optical AP, therefore the coordination among the optical APs is necessary. A third scheme using blind interference alignment (BIA) with RS is introduced to address these challenges. In BIA-RS, users are spatially divided into different groups, where RS manages intra-group interference and BIA offers coordination among multiple optical APs and allocates non-orthogonal resources to all groups with guaranteed inter-group interference cancellation. To further enhance the performance of this scheme, an optimization problem is formulated for power allocation to maximize the minimum sum rate within each group. The proposed BIA-RS scheme provides higher sum rates compared to benchmarking schemes such as BIA, RS, and NOMA.
Finally, in OWC, power allocation optimization problems are complex. Therefore, deep neural network (DNN) models are introduced to obtain real-time solutions with low computational complexity. The results show the effectiveness of the trained DNN in enhancing the performances of the proposed transmission schemes, HRS and BIA-RS
Resource Allocation in Cellular Optical Wireless Systems Using Reinforcement Learning
No full textIn modern communication networks, indoor users and their data rate demands are massively increasing. Radio-based systems struggle to meet these demands due to their limited available spectrum. Researchers proposed optical wireless communication (OWC) systems in indoor environments to meet these demands due to their high data rates, reliability and energy efficiency. As the number of users increases, efficient allocation of system resources becomes essential.
The thesis introduces, for the first time a Q-Learning (QL)-based resource allocation algorithm tailored for indoor OWC systems. The proposed method was evaluated on wavelength division multiple access (WDMA)-based visible light communication (VLC) and steerable laser-based OWC systems. The QL approach delivers resource allocation solutions comparable to the optimal solutions achieved by mixed integer linear programming (MILP), while operating without prior environmental knowledge. Nevertheless, the method’s reliance on discrete Q-tables limits its applicability to small-scale environments with fewer users and access points.
To address more complex scenarios, the thesis proposes a deep reinforcement learning (DRL) framework for resource allocation in indoor OWC systems. The DRL approach uses neural networks to replace the Q-table, enabling operation in larger, more dynamic settings. Evaluations show DRL achieves performance close to MILP benchmarks and outperforms simpler heuristics, such as distance-based allocation (DBA). However, its effectiveness depends on careful hyperparameter tuning and model design.
Finally, the thesis integrates artificial neural network (ANN)-based user positioning with DRL to manage resource allocation under user mobility. Using a
V
random waypoint (RWP) mobility model, the hybrid ANN-DRL system significantly improves learning performance in dynamic conditions. Yet, the system assumes a specific room configuration; environmental changes, such as moving furniture or introducing obstacles, may reduce positioning accuracy and require periodic retraining to sustain performance
Eco-friendly marketing strategy and performance outcome: the role of learning
No full textAmid growing concerns about environmental issues that focus on bringing the ecosystem back to its full functionality, firms must develop and utilize eco-friendly marketing competency to achieve performance outcomes, including marketing, financial, and environmental performance indicators. Despite the substantial research and tremendous development in eco-friendly marketing practices, empirical research on eco-friendly marketing strategies that examines the complexities of eco-friendly marketing strategies and the relationship between performance outcomes remains inconclusive. Yet little is known about the process of the performance impact of an eco-friendly marketing strategy. Consequently, marketing managers lack the guidance to understand the complexities and nature of the implications of an eco-friendly marketing strategy. Many scholars also argue that an implemented marketing strategy can generate new knowledge, utilize existing knowledge, and reduce the anxiety for developing marketing competency. Therefore, this thesis examined how eco-friendly exploratory learning, eco-friendly exploitative learning, and eco-friendly learning anxiety as eco-friendly marketing competencies can enhance the performance outcome of manufacturing firms. Drawing on the achievement goal theory and stakeholder theory, a comprehensive model outlining the underlying mechanism in the relationship between eco-friendly marketing strategy and performance outcomes and the effectiveness of eco-friendly marketing strategy for the mechanisms was developed. A quasi-longitudinal survey design with 3 waves of data collection from 296 manufacturing firms was used to test the model. The direct, mediating, and moderating effects were examined using structural equation modelling. The results support the positive relationship of eco-friendly marketing strategy with eco-friendly exploratory learning and eco-friendly exploitative learning. Further, the study also shows a negative relationship between eco-friendly marketing strategy and eco-friendly learning anxiety. Eco-friendly marketing competencies are found to impact performance outcomes positively. Further, the study also provides evidence for the mediating effects of eco-friendly exploratory learning, eco-friendly exploitative learning, and eco-friendly learning anxiety. Results also show that competitive intensity and customers’ environmental sensitivity negatively moderate the eco-friendly marketing strategy-eco-friendly marketing competency link, whereas coordination flexibility and ties with intermediaries moderate the link positively. These findings have important implications for marketing theory and practice. Stating the limitations of the study, future research avenues are also highlighted
Deep Learning-Based Automatic Segmentation of Skeletal Muscles and Generalisation Study
No full textMusculoskeletal disorders affect a significant portion of the population, with one in four people in the UK currently suffering from such conditions, which impact both individuals’ work and social lives. While the mechanisms underpinning skeletal disorders are well-understood, a major challenge in advancing the understanding of muscle-related conditions lies in the difficulty of accurately measuring muscle tissue's physiological status.
Muscle disorders vary significantly in terms of their causes, affected muscles, progression rates, and treatment strategies. Furthermore, individuals with the same muscle disorder often exhibit different responses to the condition, highlighting the need for subject-specific, quantitative characterisation of muscle tissue in vivo. This could significantly enhance current diagnosis and treatment strategies and provide a more informed approach to evaluating the efficacy of new treatments in clinical trials. However, despite its potential, quantitative muscle tissue analysis has not yet been fully integrated into clinical practice. As an indispensable part of this quantitative analysis, manual segmentation of muscle images is labour-intensive, time-consuming, prone to inter- and intra-operator variability, highlighting the need for automatic segmentation methods.
The aim of this thesis was to develop, test, and analyse methods for deep learning based automatic muscle segmentation from medical imaging data. This work presents three distinct methods designed to address the limitations of the current method for muscle automatic segmentation in MR images. The outcome provides a comprehensive overview of both existing and novel methods for muscle segmentation pipeline and analysis from medical images.
The methods discussed in this thesis offer valuable insights for future research, providing a foundation for the quantitative study of muscle segmentation. By adopting the best-suited deep learning models or pre-/post-processing pipeline from this work, future studies can improve the understanding and treatment of muscle conditions. Ultimately, this research aims to promote the clinical adoption of computational tools for muscle disorder characterisation, enhancing diagnosis, treatment planning, and patient monitoring
Deep Learning-based Error Level Modelling for Image Manipulation Detection and Localization
No full textThe widespread accessibility of multimedia content through social platforms has significantly increased the risk of image manipulation, thereby amplifying the dissemination and influence of fabricated visual information. Addressing this critical challenge necessitates the development of robust, efficient, and explainable methods to verify the authenticity of visual content. This thesis investigates novel AI-based approaches for image manipulation detection, with a particular emphasis on achieving strong generalization across challenging datasets and enhancing model interpretability.
Specifically, three deep learning architectures are proposed to address different tasks within image manipulation detection and localization: 1) WCBnet introduces an adaptive cross-block weighting mechanism at the convolutional block level, allowing the network to dynamically fuse low-level and high-level features based on their relevance to manipulation cues. This hierarchical feature weighting strategy enables WCBnet to achieve fast and accurate fact-checking with minimal additional computational overhead (only a 2.3% increase in trainable parameters), making it particularly suitable for large-scale scenarios involving newly generated manipulated images. 2) DenseWCBnet extends WCBnet by integrating multi-scale receptive fields across multiple convolutional blocks. Through adaptive fusion across different spatial dimensions, DenseWCBnet generates densely weighted feature representations that further enhance manipulation detection accuracy and significantly improve robustness against diverse and challenging manipulation types. 3) WSWCBnet proposes a novel weakly supervised localization framework, combining image-level manipulation heatmaps and semantically irrelevant segmented maps to localize manipulated regions without the need for pixel-level annotations. By leveraging only image-level supervision, WSWCBnet achieves pixel-level localization performance comparable to fully supervised methods, substantially reducing the annotation burden while maintaining high localization precision.
Extensive experiments demonstrate that WCBnet and DenseWCBnet consistently outperform state-of-the-art methods across six widely-used datasets in terms of classification accuracy and F1-score. Furthermore, when evaluated on the particularly challenging DeepfakeArt generative dataset, WCBnet and DenseWCBnet achieve classification accuracies of 94% and 97%, respectively. Additionally, WSWCBnet demonstrates its effectiveness in manipulation localization, achieving comparable performance to fully supervised models despite relying solely on weak supervision. Overall, this thesis advances the field of image forensics by providing robust, explainable, and annotation-efficient deep learning solutions for image manipulation detection and localization, with strong applicability to real-world forensic scenarios
Experimental study of ammonia-methane diffusion flames
No full textAmmonia could either be mixed with or replace conventional fuels in combustion processes to provide a low or zero-carbon solution for these systems. Several studies of ammonia-methane combustion have been reported in the literature. However, the literature still needs to include experimental data on the combined effects of ammonia and nitrogen doping on methane-air diffusion flames to delineate the chemical impact of fuel-bound nitrogen and the thermal/dilution effect of free nitrogen on ammonia-methane-air combustion. Therefore, this study aims to provide insight into the properties of ammonia/methane combustion in the presence of nitrogen using various flame configurations under laboratory conditions. This work investigates a range of ammonia/methane/nitrogen mixtures in three combustion rigs: laminar counterflow diffusion burner, co-flow diffusion burner, and swirl-stabilised co-flow diffusion burner. The experimental campaign included temperature measurement using fine wire thermocouple and planar laser-induced fluorescence studies to provide NO and OH species concentration profiles through the flames and a spectroscopic method of temperature measurement to compare to the thermocouple measurements. The recorded LIF signals were corrected for quenching and Boltzmann population fraction to quantify the OH and NO concentrations. Preliminary tests and analyses were conducted to select a suitable transition pair for the OH PLIF thermometry.
Furthermore, chemiluminescence visualisation studies were conducted to record the OH* and CH* species in the flames. A flame stability test was also performed to establish the flame stabilisation region and the blow-off transient behaviour. The corrected thermocouple temperature measurement reproduced the OH PLIF temperature results excellently. Increasing the strain rate of the counterflow diffusion flame extended the detectable temperature range. This study contributes to the experimental database of ammonia/methane combustion in the presence of nitrogen. The results revealed opposite effects of nitrogen addition on the temperature and OH LIF intensity in the mid-section and tip of the co-flow flame. The CH* Chemiluminescence profiles of the co-flow flame featured an inflexion point between the peak and vanishing points. The OH* chemiluminescence did not feature any inflexion points. Nitrogen addition to the ammonia-containing flames suppressed the NO emissions in the co-flow flames. Primary and secondary NO formation zones in the counterflow ammonia-methane-air flames are reported for the first time. The data presented in this work facilitates the chemical kinetic and CFD modelling of ammonia/methane diffusion flames. Further studies to determine the temperature of the swirling co-flow ammonia-methane-air diffusion flame using two lasers to record the OH PLIF intensity at a transition pair simultaneously and PIV imaging to capture the flow field are suggested
γ-cyclodextrin metal-organic frameworks for controlled release
No full textThis thesis explored the use of γ-cyclodextrin metal-organic frameworks (γ-CD-MOFs) as a controlled-release delivery method for curcumin, a bioactive compound with significant pharmacological properties but low bioavailability due to low solubility and rapid degradation. The research aims to enhance the solubility, stability, and bioaccessibility of curcumin by utilizing the unique structural functional features of γ-CD-MOFs, which combine the porous nature of metal-organic frameworks with the biocompatibility of cyclodextrins.
The synthesis of γ-CD-MOFs was investigated after a thorough review that highlighted its distinct structural characteristics, biocompatibility, and potential as delivery vehicles for bioactive compounds. This demonstrated that the choice of solvent had a major impact on crystal morphology and that the porosity was essential to the functional performance of γ-CD-MOFs. Further investigation of the interaction between curcumin and γ-CD-MOFs indicated that curcumin may be present on crystal surfaces or between γ-CD pairs, in addition to the hydrophobic cavities of γ-CD-MOFs, affecting its release profile. Investigations into the synergistic effects of γ-CD-MOFs with surfactant micelles revealed that curcumin’s apparent solubility and bioaccessibility were improved. However, when γ-CD-MOFs were incorporated into emulsion systems, it became clear that they were unstable in acidic gastric conditions. A promising delivery method for preventing γ-CD- MOFs from degrading was found to be capsule-based delivery, enabling prolonged release as well as enhanced intestinal bioaccessibility and absorption.
In summary, this thesis highlights the potential of γ-CD-MOFs as a versatile and effective delivery system for controlled release, offering a pathway to enhance the bioavailability of curcumin and potentially other poorly soluble bioactive compounds. These insights contributed to the growing body of knowledge on γ-CD-MOFs, paving the way for their application in pharmaceutical and functional food formulations
Exploring Research, Policy and Practice in Early Literacy: Teach me where I am
Get PDFAbstract
There is evidence that a complex ‘gap’ exists between early literacy academic research and practice (Vanderlinde and Van Braak, 2010). Boundaries between research, policy and practice have become disjointed due to policymakers’ increasing intervention in the early years curriculum, and the focus on synthetic phonics programmes (Ellis and Moss, 2014). Historic and current ‘reading wars’ have resulted in phonics now taking prime place in policy with the Science of Reading (SoR) movement being used in the public debate to advocate policies and instructional approaches that draw evidence largely from the cognitive perspective and the cognitive processes involved in reading (Shanahan, 2020). The most recent debate centres around how to meet the needs of individual children (Wolf, 2015) to address the ongoing problem of underachievement and issues of equality (Hall, 2003).
Framing the study around the sociopolitics of evidence-based practice (Clegg, 2005), and viewing early literacy through four perspectives (cognitive, psycholinguistic, socio-cultural and socio-political), this study aims to explore the complex relationship between research, policy and practice in relation to the teaching and learning of early literacy in classrooms. Specifically, it aims to gain insight into classroom practice and how far that practice relates to academic educational research and theories of learning. The study also captures the perspectives of children to further understand how they experience literacy in classrooms.
A case study was conducted using a mixed-method approach consisting of both qualitative and quantitative methods. The research took place in a Multi Academy Trust in the north of England where I am employed as a Teacher and Family Literacy Project Lead. 10 Participants took part in the study. A policy analysis of The Reading Framework (DfE, 2021) was conducted using Hyatt’s (2013) Critical Policy Discourse Analysis Frame (CPDAF) in order to explore the evidence-base for this key policy in early literacy teaching. Semi-structured interviews were conducted with 4 teachers and structured activities were conducted with 6 children, aged 3 to 7 years. Lesson observations were conducted of reading, phonics and writing lessons and the 6 children were observed in these lessons. Children’s emergent literacy skills and knowledge were assessed using a number of assessment tools. Marking and feedback, and learning objectives were analysed in children’s books.
Thematic analysis (Braun and Clarke, 2006) was used to analyse data with 4 main themes and sub themes evident in the data set: phonics, teachers’ practices, teachers’ perspectives and children’s literacy experiences. The findings indicate that in policy and practice, there is a heavy focus on the teaching of phonics; therefore, there is a need for a more research-based balanced approach to early literacy teaching, and specifically, for developmentally driven instructional approaches (Bear et al., 2020) which will better meet the needs of all children. The study discusses implications for research and practice and makes recommendations for research to help bridge the research-practice gap as well as making recommendations for teachers’ early literacy practices (Paechter, 2003)
Tuneable liquid crystal elastomer devices with electric-field induced alignment
Get PDFAs a class of stimuli-responsive materials, liquid crystal elastomers (LCEs) have attracted increasing attention due to their ability to translate microscale orientational changes into macroscale deformation. Considerable efforts have been devoted to controlling the molecular alignment of LCEs to enable their functionality across various applications. Among them, intrinsic auxetic liquid crystal elastomers (IALCEs) represent an emerging system, where the regulation of auxetic responses through alignment control remains unexplored. Moreover, there is significant potential for designing practical applications based on their tuneable auxetic responses. This thesis aims to achieve orientation control of IALCEs via electric fields, by introducing the Fréedericksz transition alignment, thereby enabling tuneable auxetic responses in IALCEs polymerised under electric fields. Furthermore, by patterning distinct auxetic responses, this work demonstrates a pathway toward tuneable and functional implementations of IALCEs, with particular emphasis on applications such as information encryption.
Due to the elastic and dielectric properties of liquid crystals, the director can deform under an external electric field. In this thesis, we conduct electro-optical experiments to characterise the threshold voltage of the IALCE precursor, which was determined to be 0.9 ± 0.1 Vrms. Together with the measured elastic constants (K11 = 4.6 pN and K33 = 4.3 pN), we construct a detailed schematic illustrating the relationship between applied voltages and the director profile under Fréedericksz transition alignment in the IALCE precursor. Through applying electric fields during polymerisations, we represent the first successful attempt to obtain high-quality homeotropic alignment in 100 μm thick films of IALCEs. We also fabricate planar-aligned samples for comparison. Both types of LCE sample show an auxetic response with threshold strains in excellent agreement: 0.56 ± 0.05 for the homeotropic alignment sample and 0.58 ± 0.05 for the planar alignment sample. Furthermore, we demonstrate that the system became biaxial even at very low strains with the high-quality homeotropic sample. To explore the tunability of auxetic responses, we introduce Fréedericksz transition alignments by polymerising the IALCE precursor under varying electric field strengths. By adjusting the applied voltages, we fabricate IALCEs with distinct director profiles, exhibiting the maximum angle induced in the Fréedericksz transition alignment from 40° to 88°, and realise a wide tuneable range of threshold strain of auxetic responses from 0.58 ± 0.05 to 0.91 ± 0.05. Thermal deformation and birefringence measurements confirm the successful introduction of the Fréedericksz transition alignment, showing that IALCEs with the Fréedericksz transition alignment exhibit minimised actuations due to modified order parameters. DSC and stress-strain curve measurements confirm consistent glass transition temperatures (15.0 ± 1.0 °C) and hyperelastic properties across all samples, indicating that the tunability of the auxetic response originates solely from variations in the director alignment.
To explore the potential of tunable devices based on IALCEs, we fabricate two demonstrations of multilevel, multidimensional information storage and encryption by integrating electric field-assisted polymerisation with photomask patterning. These devices respectively showcase 2D optical and 3D tactile information encoding. For the 2D optical information, binary codes (000 100 110), (011 001 000), and (100 010 001) are encrypted into a single IALCE film via masked polymerisation under electric fields. The encoded patterns are sequentially decrypted by stretching the film to specific strain levels of approximately 0.60, 0.70, and 0.80, respectively. In the case of 3D tactile information, we encode the haptic letters "K", "O", and "R" using a set of four IALCE films. Upon sequential stretching to strains of approximately 0.75, 0.85, and 1.00, the corresponding haptic letters appear in order. Owing to the intrinsic auxetic response of the material, a fourth haptic letter, "W", emerges on the reverse side of the IALCE set, completing the word "WORK" during the final stage of decryption. The decryption processes for the 2D optical and 3D tactile information remain mutually independent and highly secure, as successful decryption relies not only on the strain magnitude but also on the strain direction.
In summary, this work demonstrates that polymerisation under electric fields enables precise control over the director profile in IALCE precursors, thereby tuning the auxetic responses of polymerised IALCEs. The successful fabrication of high-quality homeotropic alignment confirms that the emergence of biaxiality is an intrinsic feature of the auxetic response in nematic LCEs. What’s more, by patterning distinct auxetic responses, we realise multilevel, multidimensional information storage and encryption, highlighting the potential of IALCEs in tuneable information devices. This study offers a new strategy for designing and implementing tuneable devices based on programmable auxetic responses in IALCEs
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇