70040 research outputs found
Sort by
Establishing Green Infrastructure Planning Framework based on the Perspective of Environmental Efficiency – A case study of Taipei Basin, Taiwan
No full textClimate change brings significant risks for urban environments, for example flooding and extreme heat events. This makes the implementation of green infrastructure crucial for enhancing resilience and addressing climate-related challenges. Despite the rapid growth of policy initiatives on green infrastructure planning, urban resources, such as land, financial, and human capital, are limited, highlighting the need to use these resources efficiently and wisely. Therefore, it is essential to plan green infrastructure that maximises multiple ecosystem services at minimal cost, aligning with the concept of environmental efficiency.
This thesis proposes a green infrastructure spatial planning framework based on the perspective of environmental efficiency. Using the Taipei Basin as a case study, it provides a comprehensive exploration of green infrastructure siting. The first section clarifies the role of green infrastructure in achieving Sustainable Development Goals, emphasising the importance of environmental efficiency in green infrastructure planning, as revealed through a bibliographic analysis of scientific literature. In the second section, green infrastructure priority mapping assists decision-makers in identifying priority areas for green infrastructure siting to maximise benefits. Subsequently, the potential green infrastructure locations are identified based on the priority map and their costs and improvements are evaluated across multiple ecosystem services. Data envelopment analysis is then employed to assess the environmental efficiency of each potential location. In the final section, data envelopment analysis is integrated with artificial intelligence algorithms to develop DEA-ML models and employs explainable artificial intelligence to interpret the best performance model.
The green infrastructure spatial planning framework serves as a planning support system to identify priority areas and to measure and predict the environmental efficiency of green infrastructure. The framework offers valuable insights for decision-makers to strategically site green infrastructure, fostering urban resilience and contributing to the advancement of multiple Sustainable Development Goals
Leaderisation: How leadership thinking changed the way knowledge is produced about schools
No full textThe present study poses the issue of ‘leaderisation.’ For two decades, school knowledge production has been framed in terms of leadership inquiry. Research has trended towards ‘leader-centrism,’ and policy has undergone a ‘leaderist’ turn. However, this trend in knowledge has not yet been the target of a dedicated investigation. The present study explores this trend, tracing how leadership inquiry diffused and reproduced across school knowledge. The study’s historical and philosophical approach draws on concepts and methods from Michel Foucault’s history of systems of thought. It conceptualises leaderisation as a leadership ‘thought-system’ reframing school ‘discourse.’ The study accounts for how school discourse and leadership inquiry became a fit, over four steps. Firstly, leadership inquiry was introduced into 1950s school discourse. However, it did not diffuse or reproduce yet, because of its weak affinity for the administrative research favoured at the time. Second, leadership inquiry underwent rethinking in the 1970s, with the realisation that its individualist focus had a weak affinity for organisation-level analysis. It was also recognised that there was a weak synergy between ‘what’ leadership inquiry was researching and ‘how’ it was researching it. Thirdly, leadership inquiry came into its own in the 1970s, with a new thought-system built on a synergy from investigating ‘surveyable leadership’ through ‘leadership surveys.’ Fourth, this new thought-system was imported into 1980s school discourse for two reasons. On the one hand, the synergy between leadership and surveys displaced concerns about how to integrate and build upon knowledge. On the other hand, there was a strong affinity between leadership inquiry and school discourse, as they shared an interest in how leaders articulate a common purpose. From there, leadership inquiry would diffuse by incorporating questions about school systems, students, other countries, and stakeholders. Leadership research would reproduce through theoretical consolidation, and the aggregation of past research in meta-reviews. Overall, the present study argues that leaderisation has stabilised school knowledge production, but left it overly concentrated. This presents an opportunity for scholars to diversify, by pursuing diverging research programmes. The study concludes by indicating directions for further historical and empirical research, as well as for strengthening research design
A change of heart: the effect of high flow arteriovenous fistulas on cardiovascular outcomes - a systematic review and synthesis without meta-analysis (SWiM)
No full textBackground and hypothesisAlthough arteriovenous fistulas (AVF) are widely accepted as the gold standard form of dialysis access for haemodialysis patients, high AVF flow may impose significant haemodynamic stress, potentially contributing to adverse cardiovascular events, including heart failure (HF). We hypothesize that higher AVF flow is associated with adverse cardiovascular outcomes including an increased incidence of heart failure. We sought to clarify this relationship as understanding this association is crucial for improving patient care.MethodsA comprehensive search of the PubMed, EMBASE, and Cochrane databases was used to identify observational studies and randomized controlled trials reporting an effect of AVF flow on cardiac outcomes (clinical, echocardiographic, and biomarker). Due to study heterogeneity, meta-analysis was not feasible. Synthesis without meta-analysis (SWiM) was performed using vote counting of direction of effect as the primary outcome.ResultsHigher AVF flow rates were consistently associated with increased incidence of HF and worsening HF symptoms. Cardiac imaging revealed left ventricular dilation and reduced left ventricular ejection fraction in patients with high-flow AVFs. Elevated biomarkers, such as natriuretic peptides, corroborated the adverse cardiovascular effects of high AVF flow.ConclusionsThis systematic review and synthesis without meta-analysis showed a positive relationship between AVF flow and clinical, echocardiographic, and biomarker cardiovascular outcomes. The methodological heterogeneity of studies highlights the need for well-designed prospective research with standardised definitions of high flow AVFs and measures for reporting of cardiovascular outcomes
Machine Learning-Based Design Systems for Holistic Landscape Integration of Traditional Settlements: Evolutionary Models Applied at Vikos Gorge
No full textThe advancement of technology implies a simultaneous evolution of design tools, particularly in architecture and in the field of urban design. A new challenge has emerged, enabling a more holistic approach that considers the balance between environmental conditions and anthropogenic factors influencing the design process. This shift, coupled with the integration of Machine Learning systems that are continuously receiving feedback, is reshaping core design principles, moving away from a human-centered focus. The purpose of this research is to explore the innovative perspectives that such technological advancements offer, analyzing the dynamic relationships between the human subject and the natural elements. The goal is to develop a design system that can adapt to respective fields of study and anticipate changes based on input data. The traditional architecture initiated a dialog with contemporary culture through the application of emerging computational tools
Efficacy of Novel Antimicrobials against Viruses
No full textThe recurring outbreaks of viral diseases, emphasized by the COVID-19 pandemic, have exposed significant gaps in antiviral therapeutic strategies, underscoring the urgent need for innovative and effective solutions. This thesis investigated the antiviral efficacy and mechanisms of cationic natural peptides, synthetic peptides, and novel anthranilamide-based peptide mimics against a range of enveloped and non-enveloped viruses. The enveloped viruses studied were the coronavirus murine hepatitis virus 1 (MHV-1; a surrogate for SARS-CoV-2), influenza virus A strains (H1N1 and H3N2), and Herpes simplex virus 1 (HSV-1), while the non-enveloped viruses were human adenovirus 5 (HAdV-5) and murine norovirus 1 (MNV-1).
The antiviral activity of these compounds was assessed through direct virus treatment, host cell pre-treatment, and post-infection treatment. The compounds acted by directly targeting viral structures, with no indirect effects on host cells. Peptide mimics had strong antiviral activity against enveloped viruses but were ineffective against non-enveloped viruses. Among them, RK610 and RK758 exhibited potent efficacy, with RK610 showing low IC50 values against MHV-1 (2.38 µM), H1N1 (2.35 µM), H3N2 (3.7 µM), and HSV-1 (34.9 µM). Similarly, RK758 displayed notable activity against MHV-1 (23.7 µM), H1N1 (12.34 µM), H3N2 (9.3 µM), and HSV-1 (24.9 µM). In contrast, both natural and synthetic peptides had antiviral activity against non-enveloped viruses. Notably, Mel4 demonstrated significant activity against norovirus (IC50 = 8.6 µM) and moderate activity against adenovirus (IC50 = 47.43 µM), further highlighting its broad-spectrum potential. Transmission electron microscopy revealed that peptide mimics disrupt viral lipid envelopes, likely by interacting with and destabilizing the lipid bilayer, while peptides targeted viral capsids in non-enveloped viruses.
To explore the key factors influencing antiviral activity, a structure-activity relationship analysis was conducted by examining the effects of modifying hydrophobic and cationic groups. The varying efficacy against different virus groups is likely driven by factors such as charge, hydrophobicity, molecular size, and, in the case of natural and synthetic antimicrobial peptides, their secondary structure. For instance, RK610 (491.60 Da weight, +2 charge) was hydrophobic with a naphthyl group, while RK758 (759.71 Da weight, +2 charge) contained a guanidine cationic group and a tryptophan residue. Mel4, with a molecular mass of 2347.8 Da and a +14 charge, was hydrophilic. Circular dichroism (CD) spectroscopy showed that Mel4 adopted a random coil in aqueous solutions but forms α-helices in sodium dodecyl sulfate, whereas RK758 exhibited a slight β-sheet structure in both conditions, suggesting a different antiviral mechanism. Most of the compounds tested were not toxic to the viral host cells. Two peptide mimics (RK610 and RK758) and two peptides (Mel4 and melimine) were selected for further investigation. Checkerboard assays demonstrated strong synergy, with RK610+RK758 having synergy against MHV-1 and H1N1 (∑FICI values of 0.14 and 0.5, respectively), while 610+Mel4 showed potent synergy against HSV-1 (∑FICI = 0.18). These combinations were also not cytotoxic.
Further biophysical analyses using 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC, 100%) and 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC): 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS) (70:30) lipid membranes provided mechanistic insights into peptide-mimic interactions with lipid envelopes. Tethered bilayer lipid membrane (tBLM) assays revealed that both peptide mimics and their combinations reduced conductance, regardless of lipid composition. Quartz crystal microbalance with dissipation monitoring (QCM-D) showed that RK610 and RK758 induced mass addition at the outer layer, with frequency shifts of -5.6 Hz and -6.5 Hz at the 3rd overtone, respectively. These shifts reflected an increase in mass at the surface layer, with minimal and linear changes in dissipation, suggesting relatively rigid binding. In the presence of anionic lipid POPS, mass attachment increased significantly (p <0.0001), with frequency shifts of -14.5 Hz (RK610) and -12 Hz (RK758). Dissipation changes became more pronounced and non-linear, indicating altered viscoelastic properties. Combinations of these mimics further enhanced mass addition and rigidity changes, particularly with anionic lipids (p <0.0001). Langmuir-Blodgett surface pressure analyses revealed distinct behaviours in DOPC and DOPC+POPS monolayers. In DOPC, RK610 increased the pressure to 44mN/m, while RK758 reduced it to 36mN/m. The RK610+RK758 and RK610+Mel4 combinations raised maximum pressures to 44mN/m and 45mN/m, respectively. In DOPC+POPS, RK758 destabilized the monolayer (35mN/m, no plateau), whereas combinations restored stability.
This research also explored the application of these compounds as antiviral surface coatings to inhibit the indirect transmission of viruses from inanimate surfaces. Polydopamine was used as a functional coating material to immobilize the cationic peptide Mel4 and mimic RK758 on glass coverslips. Surface characterization using X-ray photoelectron spectroscopy (XPS), time-of-flight secondary ion mass spectrometry (TOF-SIMS), and water contact angle analysis confirmed successful coating and revealed distinct chemical and physical properties of the modified surfaces. XPS revealed changes in the surface elemental composition following treatment with peptide and mimic, notably an increase in nitrogen (N) and carbon (C) signals, consistent with the presence of peptide/mimic layers. ToF-SIMS further confirmed the successful surface coating by detecting characteristic ionic fragments corresponding to amino acid residues, such as those containing arginine and tryptophan. Additionally, water contact angle analysis demonstrated changes in surface wettability, reflecting the hydrophobic or hydrophilic nature of the surface after peptide/mimic coating.
Antiviral assays indicated that RK758-coatings reduced the infectivity of MHV-1 and influenza H1N1 by 68% and 89%, respectively. In contrast, Mel4 coatings reduced the infectivity of the non-enveloped human adenovirus type 5 (HAdV-5) and murine norovirus (MNV-1) by 85% and 90%, respectively. Durability tests over 74 hours revealed that RK758-coated coverslips maintained ~50% activity, while Mel4-coated surfaces maintained >80% activity. Biocompatibility assays confirmed safety profile of the coated surfaces, with host cells (9, Vero, MDCK, and Raw 246.7) viability exceeding 90%.
In summary, this thesis highlights the potential of cationic peptides and anthranilamide-based mimics as broad-spectrum antiviral agents, both as therapeutic formulations and as surface coatings. By elucidating their mechanisms of action, structure-activity relationships, and synergistic effects, this work provides a robust framework for the development of novel antiviral strategies to combat both enveloped and non-enveloped viruses
Securing Users' Data After Death
No full textIn the digital age, the online data accumulated by individuals often persists after death, but the current mainstream platforms lack systematic technical support and institutional arrangements for posthumous data management. Moreover, Terms of Service (ToS) governing data inheritance and account handling are typically vague, complicating users' efforts to manage their digital legacies. This thesis focuses on the two major issues of "security of data inheritance" and "interpretability of service agreements", and proposes and implements two complementary systems: BeyondLife and TOSense.
BeyondLife is the first cross-platform, open-source, and privacy-enhancing digital will management solution designed to securely handle and distribute digital assets after death. Built upon a novel encryption scheme introduced in this thesis - Partially Decryptable Ciphertext Policy Attribute-Based Encryption (PD-CP-ABE) scheme, it allows heirs to access partial content according to their associated attribute without repeated encryption. The correctness and confidentiality guarantees of PD-CP-ABE have been formally verified using the Tamarin Prover. Unlike existing systems, BeyondLife operates independently of service providers, giving users complete control over how their digital wills are created, stored, and executed. To ensure the transparency and integrity of the execution process, BeyondLife uses blockchain to record the triggering and execution of the will.
TOSense is an interactive browser plug-in designed to help users understand the privacy-related content in the ToS. The system employs a customised lightweight crawler to retrieve and clean ToS content in real time, and then identifies relevant sections and generates concise, user-facing summaries through semantic matching and large language models (LLMs). To assess reliability without manual annotation, TOSense introduces a Question/Answer Evaluation Pipeline (QEP) that automatically evaluates answer correctness through clustering and topic matching, improving ToS readability and users’ comprehension of data-usage rules.
Together, the two systems build a trusted data governance architecture that spans from "pre-death knowledge" to "posthumous execution", providing both theoretical support and practical pathways for the design of verifiable digital legacy systems. Both are publicly available: BeyondLife supports iOS TestFlight Beta and simulator deployment, and TOSense operates as a Google Chrome extension
Advancing the Assessment and Encouragement of Myocardial Recovery in LVAD Patients
No full textCardiovascular disease, including heart failure with reduced ejection fraction (HFrEF), remains the leading cause of morbidity and mortality, exacerbated by a chronic shortage of donor hearts. Left Ventricular Assist Devices (LVADs) function as implanted electromechanical pumps that restore systemic blood flow, typically as a bridge to transplant or destination. However, LVADs have also shown the potential for myocardial recovery (MR), a phenomenon characterised by LVAD-induced reverse remodelling of the failing heart, which, if sufficient, can allow mechanical circulatory support to be discontinued. MR offers long-term outcomes comparable to or better than transplant or indefinite LVAD support while avoiding the complications of both, however the rates of complete MR remain relatively low and inconsistent despite being the ideal outcome.
This thesis addresses this by advancing the assessment of MR in LVAD-supported HF patients, specifically by developing a novel non-invasive diagnostic index, derived from LVAD pump data, for assessing native cardiac function (NCF) and so quantifying MR in LVAD patients. Existing pump-derived indices in the literature were first evaluated in a computational simulation environment to identify limitations and the best performing index as a benchmark. This was necessary since no common protocol existed among these studies to evaluate MR indices. This analysis revealed a prohibitively high sensitivity to changes in preload and afterload for all indices, rendering them non-viable for clinical application.
The thesis' formulated a novel MR index, JnV, a function of pump flow and the LVADs hydraulic power, and evaluated it alongside 69 other potential indices and the literature benchmark in the same simulation environment. Cardiac function and pump speed were varied, as well as preload, afterload, and heart rate (HR) to assess sensitivity of the indices to each, encompassing 1596 different patient conditions. JnV exhibited a significant improvement in performance, with substantial reductions in sensitivity to loading conditions and HR compared to the benchmark and all tested indices, demonstrating a novel robustness under a diverse range of patient conditions while maintaining a high accuracy. This process also involved devising and providing the field with a first-of-its-kind standardised, systematic and comprehensive protocol for evaluating MR indices.
A hybrid mock circulation loop was also constructed with which in vitro data was collected to construct a new mathematical model of the HeartMate 3 (HM3) LVAD. To further validate JnV, all the indices were subject to the same in silico evaluation but with the HM3 model in place of the HeartWare VAD model used prior to ascertain whether the proposed index performed adequately across different LVAD platforms. Again, JnV demonstrated the greatest independence to cardiac loading and HR among all tested indices while providing accurate assessments of patient MR.
Current clinical gold-standards for NCF assessment in LVAD patients are invasive, require a clinical visit, interrupt LVAD operation to minimize the pump's interference of true NCF, and are time-consuming and costly as a result. In contrast, JnV, being derived from pump-data, is non-invasive, available in real-time outside of a clinical environment without interrupting the LVAD, capable of providing a more reliable measure of true NCF underneath LVAD support, and is more time and cost-effective. These properties afford the index important and impactful applications in the continuing research of MR and in clinical practice.
A real-time measure of NCF supports clinical management of patients by providing clinicians with up-to-date information on a patients cardiac state, enabling more informed and timely decisions. This includes more accurately identifying patients who have achieved sufficient LVAD-driven MR to be weaned from support, while excluding those who have not, effectively increase rates of MR while reducing risk of relapse. Thus, JnV can encourage greater rates of MR simply by making it easier and more reliable for clinicians to pursue. Additionally, the real-time measurements of JnV can encourage MR more directly through novel therapeutic strategies. For example, by serving as feedback to an adaptive LVAD control system designed to actively encourage further LVAD-driven recovery in HF patients, an emerging concept in the field.
As such, this thesis provides a MR index that, by advancing the assessment of MR in LVAD-supported HF patients, also advances the encouragement of this preferred patient outcome through broader adoption of MR as a treatment goal and novel strategies for actively driving MR in patients. Ultimately, this contributes to better long-term outcomes for HFrEF patients
Perceptual and decisional influences in serial dependence
No full textPositive serial dependence refers to the averaging of visual information across time. This occurs for a range of visual features including orientation, wherein serial dependence manifests as a skew in the response to an oriented stimulus towards previously-seen orientations. It is unclear whether this bias occurs to and operates upon early perceptual representations of orientation, or whether prior decisions about orientation bias subsequent decisions in the absence of any perceptual changes. These two explanations make different predictions about where in the visual system the bias occurs, and how it might aid temporal
continuity and perceptual efficiency. Across four experiments, I aimed to separate perceptual and post-perceptual influences on serial dependence in orientation perception. In Experiment 1, I found that serial dependence effects were better described as a bias to previous responses than previous stimuli, and that this could not be explained by the relative recency of the inducer response. In Experiment 2, I identified stimuli that produced equivalent accuracy but divergent effects on confidence to isolate decisional uncertainty from stimulus uncertainty. Stimuli associated with lower decisional uncertainty and higher subjective confidence were no more likely to bias subsequent responses – instead, serial dependence was greatest only when the current stimulus had high uncertainty. In Experiment 3, orientation decisions associated with positive feedback were more likely to bias subsequent responses, suggesting that serial dependence can be modulated by post-perceptual information. However, as I continued to observe no effect of the quality of the previous decision on serial dependence, this post-perceptual modulation may not occur via decisional certainty. In Experiment 4 I manipulated attentional load independently of perceptual precision and found that increases of serial dependence under attentional load were driven by reduced perceptual precision of the current stimulus. Neither the perceptual precision nor attentional load of the previous trial influenced the size of the bias. Overall, the strongest driver of the effect throughout these experiments was precision in the current percept – noisy, low contrast, or otherwise poorly encoded stimuli are more likely to be biased towards previous information. Serial dependence is driven by the precision of the current percept rather than previous percepts or decisions
Assessment of Energy Conversion Associated with Brittle Rock Failure
No full textRockburst remains one of the most challenging hazards in underground excavations despite decades of research, characterised by the sudden release of energy. One limitation to understanding the mechanisms of such dynamic rock failures is quantifying the energy release process, considering energy consumption and rock ejection, as well as the associated failure modes of the brittle rock mass. Consequently, energy demands for support system design to control rockburst damage remain uncertain.
This thesis employs an integrated research methodology, including microscopic analysis, laboratory experiments, and numerical modelling across multiple scales, to evaluate energy conversion in brittle rock failure under various loading conditions. A grain-based numerical model coupled with an innovative circumferential strain-controlled loading algorithm is first developed to assess brittle rock failure, considering the effects of microstructural characteristics. The algorithm is then extended to accommodate triaxial conditions, assessing the influences of external loading conditions on energy consumption and release. The results indicate that material heterogeneity and stress concentrations from stiff minerals alter microcrack propagation and affect energy conversion. However, microcrack quantity, particularly intragranular cracks under confinement, affects energy conversion under various loading conditions. At the site scale, the brittle rock failure is further investigated in a dynamic rockburst model. By assessing energy conversion in modelled rockburst, the energy contribution of various sources is quantified, and energy demands for support system design are determined. Nearly all released strain energy stems from the energy accumulated in the rock mass near drifts due to mining activities, with about 10% converted into kinetic energy. In addition, an increased residual dilation angle, which affects brittle rock post-peak behaviour, elevates energy demands for support system design.
This thesis developed new knowledge in understanding the mechanisms of brittle rock failure and associated energy conversion across multiple scales and loading conditions. The developed circumferential strain-controlled loading algorithm, together with the improved seismically triggered strainburst model and updated energy quantification procedure, collectively facilitates the determination of energy demands for support system design, ultimately aiding in controlling rockburst damages
Analytical and Numerical Modelling of Consolidation and Strain Localisation in Unsaturated Soils
No full textUnsaturated soils are commonly encountered in geotechnical engineering practice, particularly as compacted soils in roads, embankments, dams, and retaining walls. These soils present unique challenges due to the simultaneous presence of air and water phases. The intricate interactions between the solid matrix and pore fluids significantly influence the hydro-mechanical behaviour of unsaturated soils. This makes a deep understanding of key phenomena like consolidation and strain localisation in unsaturated soils essential for predicting the performance and stability of geo-structures.
This research presents comprehensive analytical solutions to determine the pore air and pore water pressure parameters of unsaturated soils under undrained conditions. The analytical solutions are derived based on the concept of the effective stress and constitutive equations that govern volume changes in solid and fluid phases. The solutions consider the effect of hydraulic hysteresis on the effective stress parameter and the soil water characteristic curve. Validation against experimental data and discussions on the effects of the initial degree of saturation, hydraulic hysteresis, and soil compressibility on the pore pressure parameters are provided. Furthermore, a comprehensive analytical framework is introduced to analyse the evolution of the pore air and pore water pressures and the settlement of unsaturated soils throughout the consolidation process. The derivation yields two simplified diffusion-form equations, analogous to Terzaghi’s consolidation equation, to account for the flow of fluid phases within unsaturated soils. Dimensionless curves are provided to illustrate the dissipation of pore water and pore air pressures throughout the consolidation process. Comparative analysis with a fully coupled model is presented to confirm the accuracy and efficiency of the proposed approach.
In addition, this study introduces a fully coupled numerical model based on the concept of the effective stress for the analysis of strain localisation in unsaturated soils. The model incorporates a novel viscoplastic constitutive model grounded in the consistency condition, capturing the rate-dependent behaviour of geomaterials while regularising the strain localisation problem. Implemented in the multi-physics simulation software COMSOL Multiphysics, this robust framework facilitates comprehensive investigations into strain localisation in multiphase porous media. The numerical simulation results demonstrate that the proposed framework effectively mitigates mesh sensitivity issues often encountered in strain localisation modelling. Through a series of simulations, the onset and progression of strain localisation in saturated and unsaturated geomaterials are investigated under quasi-static loading conditions. The effects of permeability, strain rate, confining pressure, and drainage conditions on the strain localisation in saturated porous media are examined. Furthermore, the impact of the initial suction on strain localisation in unsaturated soils is explored, offering deeper insights into this complex phenomenon