Open Research Exeter - University of Exeter
Not a member yet
41213 research outputs found
Sort by
Promoting mental health among at-risk adolescents in Malaysia (MyHeRo): study protocol for a cluster randomised controlled trial to evaluate the effectiveness and cost-effectiveness of a school-based intervention compared with study skills condition for adolescents identified as at risk for anxiety and depression
No full textBackgroundIn Malaysia, adolescent anxiety and depression are increasing faster than ever, and rates of suicidal behaviour are rising especially among those living in deprived communities. However, Malaysia’s mental health system is currently constrained by limited workforce capacity, affecting the delivery of effective interventions. The overall aim of this trial is to use a school-based intervention to promote mental health among at-risk adolescents from low-income communities in Malaysia. Our primary aim is to evaluate the effectiveness and cost-effectiveness of a school-based intervention (“Super Skills for Life”; SSL) in reducing anxiety and depression, and in improving mental wellbeing in adolescents aged 12–14 years. We also aim to determine the characteristics of adolescents who benefit from SSL, compared to those who do not, as well as to identify contextual factors related to the successful implementation of SSL in Malaysian schools.MethodsThe design will be a two-arm, cluster randomised controlled trial comparing school-based intervention (Super Skills for Life; SSL) to study skills control condition (Study Skills Programme; SSP) using a 1:1 allocation ratio. Classrooms will be the cluster unit for randomisation. Three stratification factors will be used for randomisation: school size, classes/forms and school location (urban vs rural). The study will recruit adolescents in at least 20 secondary schools in economically deprived, rural and urban regions in Malaysia. These adolescents will be invited to complete a screening questionnaire (i.e. Depression Anxiety and Stress Scale-21; DASS-21). Based on power calculation, 428 adolescents (214 per arm) who experience moderate to severe levels of anxiety and depression on the DASS-21 will be invited to participate in the trial. Classes will be randomly allocated to SSL or SSP, with eligible adolescents from each class receiving the allocated intervention. Assessment will be conducted at screening, at pre- (i.e. baseline) and post-intervention (i.e. 2 months), and at two follow-ups (i.e. 6 and 12 months post-intervention). The primary outcomes will be a reduction in anxiety and depressive symptoms, and an improvement in mental wellbeing at 12 months post-intervention.DiscussionFindings of this trial will determine if delivering a group school-based intervention by school staff for adolescents at risk of anxiety and depression is effective and cost-effective. The findings will advance understanding of the role of school staff in the delivery of a school-based intervention and will generate new knowledge on the role of socio-cultural and other contextual factors in predicting intervention uptake and treatment outcome.Trial registrationClinicalTrials.gov NCT07138664. Registered on August 16, 2025.</p
Morphological Competence in English: Comparing Advanced L2 and L3 English Learners
No full textThis thesis compares morphological competence in advanced Mandarin-speaking second-language (L2) learners (n=63) and Uyghur-speaking third-language (L3) learners (n=40) of English, focusing on derivational word parts knowledge, morphological awareness, and processing ability. Existing research on L2 English derivational morphology, often with a goal to test cross-linguistic influence in L2 learning and processing, tends to focus on learners from typologically close first-language (L1) backgrounds (e.g., European languages, which are also rich in derivation, such as German); few studies, however, have examined English learners from typologically distinct L1s such as Mandarin (unlike English and limited in derivation), let alone the comparison between L2 and L3 learners, especially those with an agglutinative L1 rich in derivation such as Uyghur. The present research aims to address this gap by examining how L1 linguistic background may shape English derivational competence; more specifically, whether L3 Uyghur-speaking learners, as opposed to their L2 Mandarin-speaking counterparts, may benefit from L1 cross-linguistic transfer. It also aims to test how affix neutrally (i.e. whether adding a suffix may result in phonological and orthographic change to the base word) may interplay with L1 background in influencing English morphological competence.
Participants were advanced English learners/users who, after completing their undergraduate studies in China, had lived and studied or worked in English-speaking countries (e.g., the UK, the US), while maintaining active use of their respective L1. A battery of computer- and paper-based tasks was administered to measure the participants’ English morphological competence with a focus on derivation. This thesis is organised around three studies, each focused on a distinct aspect of morphological competence. Study 1 assessed the participants' word parts knowledge through the paper-based Word Parts Level Test. Study 2 assessed participants’ morphological awareness using two computer-based tasks: one focused on morphological segmentation and the other on meaning inference. Study 3 focused on their morphological processing through a masked priming experiment on derived words. For the purpose of Study 3, in addition to the L2 and L3 learners, 30 native English speakers were also included.
Each study was analysed separately to compare the English morphological competence of L2 and L3 learners. Study 1 used independent t-tests to compare word parts knowledge. No significant difference was found between the L2 and L3 groups in word parts knowledge (p = .766), likely because both groups consisted of advanced learners with extensive exposure to English. Study 2 employed 2x2 mixed ANOVAs to examine differences in morphological awareness. L3 learners significantly outperformed L2 learners in both segmentation (p = .039) and meaning inference (p <.001) tasks. Participants’ performance on non-neutral words (affixation alters spelling and sound) overall surpassed that on neutral words (affixation preserves spelling and sound) in meaning inference (p =.017). L2 learners excelled on non-neutral words (p <.001), while L3 learners performed equally well on both word types. Study 3 used a 2x3x3 mixed factorial ANOVA with post-hoc pairwise comparisons to analyse the effects of word type, stimuli, and participant group on reaction times. Neutral words demonstrated faster processing for both L2 and L3 groups than did non-neutral words (p<.001). Native English speakers, however, were unaffected by the neutrality (p = .283). Furthermore, L3 learners demonstrated marginally faster responses than did L2 learners (p = .056), although their responses were still slower than those of native English speakers.
These findings, taken together, highlight the joint impact of L1 background and word/morphological properties on English L2/L3 morphological competence. Pedagogically, they emphasise the importance of targeted approaches to teaching derivation, including explicit English affix training, for English learners with structurally distinct native languages.</p
TRACE: Trajectory Recovery with State Propagation Diffusion for Urban Mobility
No full textHigh-quality GPS trajectories are essential for location-based web services and smart city applications, including navigation, ride-sharing and delivery. However, due to low sampling rates and limited infrastructure coverage during data collection, real-world trajectories are often sparse and feature unevenly distributed location points. Recovering these trajectories into dense and continuous forms is essential but challenging, given their complex and irregular spatio-temporal patterns. In this paper, we introduce a novel diffusion model for TRAjectory rECovery named TRACE, which reconstruct dense and continuous trajectories from sparse and incomplete inputs. At the core of TRACE, we propose a State Propagation Diffusion Model (SPDM), which integrates a novel memory mechanism, so that during the denoising process, TRACE can retain and leverage intermediate results from previous steps to effectively reconstruct those hard-to-recover trajectory segments. Extensive experiments on multiple real-world datasets show that TRACE out-performs the state-of-the-art, offering >26% accuracy improvement without significant inference overhead. Our work strengthens the foundation for mobile and web-connected location services, advancing the quality and fairness of data-driven urban applications. Code is available at: https://github.com/JinmingWang/TRAC</p
Measuring cultural and ethnic diversity in research and innovation
No full textCultural and ethnic diversity is increasingly recognised as a driver of creativity and innovation in research and innovation (R&I) systems, yet the metrics employed to measure it often oversimplify or overlook critical dimensions. In this paper, we critically evaluate existing concepts, data collection strategies, and indicators, identifying three pervasive shortcomings: a lack of reflexivity about underlying assumptions and biases, insufficient attention to relative abundance, and inadequate consideration of proximity/similarity. To address these issues, we present a context-specific operationalisation of the Leinster–Cobbold (2012) framework for measuring cultural and ethnic diversity in R&I, which integrates richness, relative abundance and similarity. We demonstrate the practical utility of this index through an illustrative stress-test case study of UK university research communities using name-based inference data, revealing how it offers substantial information gain compared to conventional measures. However, the index remains sensitive to methodological choices, underscoring the need for context-specific applications and critical reflections on data limitations. We conclude by advocating careful use of multidimensional cultural and ethnic diversity metrics, thereby supporting more robust and equitable insights into R&I ecosystems when aligned with appropriate concepts, data sources, and transparent sensitivity analysis.</p
Astronomically tuned chronostratigraphy of an Upper Triassic–Middle Jurassic lacustrine succession in Southwest China
No full textThe Triassic–Jurassic transition (i.e., Late Triassic to Early Jurassic, ∼237–174 Ma, TJT) represents a critical interval in Earth's history, characterized by significant environmental perturbations both on land and in the oceans. Marine sediment records from this interval have been extensively studied, while terrestrial environmental records remain limited. In this work, we present a continuous lacustrine succession through the TJT from the Chuxiong Basin in Yunnan Province, Southwest China, based on analysis of an ∼1800-m-long core. By integration of sedimentological, paleontological, geochemical, and astronomical data, we establish a detailed geochronology spanning ∼23.5 million years from the Rhaetian (Upper Triassic) to the Aalenian (Middle Jurassic), calibrated to 405-kyr (long-eccentricity) cycles. We identify three significant negative excursions in organic carbon isotopes (i.e., at the Rhaetian–Hettangian boundary, the Sinemurian–Pliensbachian boundary, and within the lower Toarcian) that correspond to global events recorded in marine and terrestrial sections. Based on an astronomically calibrated carbon-isotope stratigraphy that conforms to other Lower Jurassic sequences, the duration of the Hettangian, Sinemurian and Pliensbachian stages are estimated to be 1.9 ± 0.3 Myr, 6.6 ± 0.3 Myr, and 9.4 ± 0.3 Myr, respectively. Our results provide insight into high-resolution climate and environmental variation during this time interval, enhancing our understanding of Early Jurassic global events and refining the Early Jurassic timescale.</p
Diffusion-free dynamics in rotating spherical shell convection driven by internal heating and cooling
No full textThe bulk properties of convection in stellar and giant planet interiors are often assumed to be independent of the molecular diffusivities, which are very small. By contrast, simulations of this process in rotating spherical shells, which are typically driven by conductive boundary heat fluxes, generally yield results that depend on the diffusivity. This makes it challenging to extrapolate these simulation results to real objects. However, laboratory models and Cartesian-box simulations suggest that diffusion-free dynamics are more readily obtained if convection is driven using prescribed internal heating and cooling instead of boundary fluxes. Here, we apply this methodology to simulations of Boussinesq, hydrodynamic rotating spherical shell convection. We find that this setup unambiguously yields diffusion-free behavior for some bulk properties of the convection, such as the radial temperature contrast and the convective heat transport. Moreover, the transition from prograde to retrograde equatorial zonal flow is diffusion free and only depends on the convective Rossby number. The diffusivity dependence of other bulk properties is regime dependent. In simulations that are rotationally constrained, the convective velocities and the strength and structure of the zonal flow are diffusion dependent, although the zonal flow appears to approach a diffusion-free state for sufficiently high supercriticality. In simulations that are either uninfluenced by rotation or influenced by rotation only at large scales, diffusion-free convective velocities and zonal flows are obtained. The result that many aspects of our idealized simulations are diffusion free has promising implications for the development of realistic stellar and giant planet convection models that can access diffusion-free regimes.</p
Explainable security investment: A Shapley value inspired metric
No full textMany existing methods can effectively find the optimal cybersecurity investment, but communicating these findings to non-technical stakeholders is a well-known cybersecurity challenge [27]. This work aims to provide additional metrics that grant further insight and justifications for an implemented cybersecurity portfolio. The Shapley value is a classic concept in cooperative game theory that quantifies the fair contribution of each player to a collective outcome. In security games, it offers a natural way to measure the contribution of individual security controls to overall defence. However, the Harsanyi dividend of combining two coalitions when taking their respective security reductions as their portfolio contribution tends to negative attributions in undesirable situations. This undermines interpretability. In this work, we propose a novel measure tailored for security games which guarantees non-negative Harsanyi dividends when combining coalitions that result in the significant increase of security. The method is grounded in the composition of two key factors that capture each control’s marginal impact. These factors offer an interpretable and fair decomposition of the overall security effectiveness. Beyond interpretability, we demonstrate how these attributions can support decision-making in cyber defence planning. To address computational scalability, we present an approximation algorithm that significantly reduces runtimes with little impact on explainability.</p
On the parallels between the Maxwell and Dirac equations
No full textExploring analogies between seemingly disparate physical systems is a powerful research strategy because it allows us to transfer insights, methods, and intuitions from one domain to another, often revealing hidden connections and sparking creative breakthroughs, as exemplified by Hamilton's seminal analogy between classical mechanics and geometric optics---a correspondence that ultimately seeded the development of quantum mechanics. In this work, in the same spirit, we explore deeply the similarity between the Maxwell and Dirac equations.
We reformulate the Dirac equation in an \emph{exact} form of the Maxwell equations, which we refer to as the ``electronic Maxwell equations.'' In this formulation, the standard free Maxwell equations emerge as a special case corresponding to vanishing longitudinal components and zero mass. The presence of a nonzero mass naturally leads to longitudinal components in the matter wave, arising as a consequence of enforcing Lorentz invariance in the electronic Maxwell equations. The physical quantities described by these electronic fields closely parallel their electromagnetic counterparts. We also make a discussion on wave components we may measure if the matter wave is actually measurable.
Likewise, we reformulate Maxwell’s equations into an \emph{exact} form of the Dirac equation, which we term the ``electromagnetic Dirac equation.'' This formulation yields the spin operator for spin‑1 particles and is capable of reproducing the Klein--Gordon equation (the wave equation for electromagnetic waves). The absence of longitudinal components characterizes the electromagnetic field as a spin‑1 field. Furthermore, we find that the Zitterbewegung of the photon corresponds to the rapidly oscillating part of the Poynting vector in classical electromagnetism.
Both the Maxwell and Dirac equations can be reformulated into closely related quaternionic equations, which directly reveal the difference in how these equations transform under Lorentz transformations. Building on this observation, we propose a possible (spin) geometry underlying both the Maxwell and Dirac equations. Within this new framework, the spacetime itself is a second--order covariant or contravariant spinor. Furthermore, the electronic field is naturally identified as a spinor (first--order spinor), whereas the electromagnetic field emerges as a second--order mixed spinor. We further express both the Maxwell and Dirac equations in terms of these spinors and extend the resulting formalism to curved spacetime.
Besides, we explore the parallels between the Feynman path integral---an alternative formalism for the quantum wave equations, and the Rayleigh--Sommerfeld diffraction integral---an alternative formalism for the optical wave equation. We apply the idea developed in the diffraction theory to establish the path integral for the Schr\"{o}dinger equation---a generalized Helmholtz equation with an additional temporal derivative term, the Klein--Gordon equation---a generalized optical wave equation with a non--vanishing mass term, and the Dirac equation---the massive Maxwell equations. Based on our results, we dicuss the flaws in the Feynman checkerboard model.
Finally, we present two illustrative applications of the analogy between the Maxwell and Dirac equations. The first explores how to assign an effective ``mass'' to the electromagnetic wave. The second examines an intriguing case of a spacetime--varying electromagnetic medium that transforms the Maxwell equations into a massless Dirac equation, with the potentials taking a special form. We derive a general solution for this equation, revealing that such materials can behave like a one--sided mirror.</p
Understanding immune priming in Pacific oysters: A multi-omics exploration of transcriptomic, epigenetic and microbiome regulation
No full textAquaculture has a fundamental role to play in global food security, given the limitations for increase of captures from global fisheries and the pressures they pose on the natural environment. Infectious disease remains an important limitation for the aquaculture industry both because it is financially costly and contributes towards poor animal welfare. For molluscs, widespread mass-mortalities of Pacific oysters (Magallana gigas) have occurred in recent years linked to infection with the bacterial pathogen Vibrio aestuarianus. Disease in farmed oysters is difficult to prevent or to treat and, in this context, immune priming has been proposed as a method to increase resistance to disease, and has been reported to be effective for viral pathogens. However, it is unknown whether an immune response can be primed in oysters to provide protection against Vibrio aestuarianus upon secondary challenge. This thesis aimed to address this knowledge gap and explored the hypothesis that epigenetic and microbiome processes are involved in sustaining long-lasting immune system memory.
In order to identify a period of DNA methylation reprogramming, and therefore a candidate window of epigenome malleability, Pacific oyster embryos were exposed to the methylation inhibitor 5-azacytidine at different temporal intervals throughout early development (Chapter 2). A period of methylation inhibitor sensitivity was observed up to approximately 11 hours post fertilisation, which is indicative of a window of methylome reprogramming. Further periods of reprogramming and/or epigenetic sensitivity may exist after this window, but their presence and precise timing was not explored.
Pacific oysters were primed with heat-inactivated Vibrio aestuarianus via 24-hour bath exposure at either the larval or young spat stage, then challenged with the non-attenuated form of the pathogen at six months old. A multi-omics approach involving RNA-Seq, 16S amplicon sequencing and whole genome bisulphite sequencing was utilised to interrogate whether immune priming resulted in long-lasting molecular alterations to the oyster transcriptome, microbiome and epigenome (Chapters 3-5). For naïve oysters, transcriptomic analysis evidenced that Vibrio aestuarianus had an immunosuppressant effect on 6-month-old oyster spat, despite high survival rates and no significant alterations in oyster microbiota suggesting spat were not susceptible to disease. Immune priming at either the larval or young spat stage resulted in lasting alterations in the oyster transcriptome, microbiome and epigenome that were sustained for up to five months after the priming took place. Lasting transcriptional alterations were identified in pathways including complement, protein modification and phagocytosis, with upregulation of these pathways suggesting faster pathogen recognition upon challenge. The oyster transcriptome, microbiome and epigenome appeared to remain malleable after metamorphosis, although larval priming appeared to be more beneficial in terms of increasing within-sample microbial diversity.
Together, this thesis demonstrates that acquired immunity can be primed in Pacific oysters against Vibrio aestuarianus at either the larval or spat stage and suggests that epigenetics may be responsible for the continued alteration of transcriptional and microbiome regulation after priming. However, it remains unclear how long this acquired pathogen memory may persist for and whether priming during a window of epigenome malleability will enhance adult survival against the virulent pathogen. Overall, this work highlights the potential to utilise sensitive windows during early life to manipulate the microbiome and epigenome in order to induce lasting memory for use in disease control within bivalve molluscs. Future research is required to clarify the precise causative mechanisms of the effects seen, whether transgenerational immunological memory can be primed against Vibrio aestuarianus, and evaluate the feasibility of this approach for building pathogen resistance within invertebrate aquaculture settings.</p
Spatially explicit global environmental footprint of China's future power system under different energy transition pathways
No full textThe supply of energy, and in particular electricity, is one of the largest drivers of global environmental change. China has been the world's largest electricity producer and carbon emitter for the past decade, generating over 9.3 PWh of electricity in 2023, which accounts for nearly one-third of global output and contributing nearly 30% of global carbon emissions. Thus, China’s decarbonising power system will shape not only its own environmental future but also that of the planet. Although the power system causes a wide range of environmental impacts at different scales, locations and timeframes, policy attention has often focused narrowly on those arising from air emissions during electricity generation (e.g., climate change and air pollution). Even forward-looking research largely limits the scope to exploring future carbon emission pathways, leaving other environmental impacts understudied. This study therefore aims to systematically quantify the rapidly evolving magnitudes and spatial patterns of the associated environmental impacts of China’s power system under an accelerating energy transition.
Spatially explicit life cycle assessment (LCA) was the key method employed to assess the environmental impacts from a full life cycle perspective. It can attach spatial information to the input and output flow as an attribute in order to locate the impacts. A practical method for implementing spatially explicit LCA was then developed, based on a critical review of methodologies and applications of relevant LCAs. This method enables the spatial hotspots of various environmental footprints along the entire supply chain to be easily identified, helping to inform tailored and effective mitigation policies and avoid environmental burden shifting across countries, sectors and impact categories. Using this method, life cycle inventory (LCI) models of China’s power system in 2020 and 2050 under two distinct transition pathways were established from a cradle-to-grave perspective, integrating international trade data, national statistics, the Ecoinvent LCI database and multiple other sources. The magnitude and spatial distribution of environmental impacts were then quantified and systematically analysed using IMPACT World+, a regionalised LCIA method widely used for location-dependent impact quantification. Furthermore, to examine how different levels of spatial detail influence hotspot identification, China’s future low-carbon power system, encompassing wind, solar and power storage, was modelled at varying spatial details for comparison.
In 2020, China generated 7.0 PWh of electricity in total. The associated Greenhouse Gas Emissions, Particulate Matter Formation, Freshwater Eutrophication, Mineral Resource Use and Land Transformation were found to be 5.5 Gt CO2 eq, 1.4 Mt PM2.5 eq, 1.8 kt PO4 P-lim eq, 6.2 Mt mineral deprived and 387 km2 arable land eq, respectively. Coal power accounted for 50% to 95% of four of these five impact categories except for Mineral Resource Use. Future renewable-dominated power systems are projected to produce 10.1 PWh and 13.1 PWh electricity under the NDC and PEAK30 scenarios, respectively. However, they are also estimated to deliver substantial synergies with air emission-related impacts, with 42%-119% reductions in GHG emissions and 31-70% reductions in Particulate Matter Formation. In contrast, the future power systems will also introduce notable resource-oriented trade-offs at the same time, including a 24-400-fold surge in Land transformation, a 0.48-5.2-fold increase in Freshwater Eutrophication, and a 0.27-1.13-fold increase in Mineral Resource Use. These environmental burdens are increasingly shifting along the upstream and downstream stages of power supply chains. For example, upstream stages include the extraction and processing of materials and the manufacturing of power technology infrastructure, whereas downstream stages involve the treatment of retired power plants and supporting facilities. As these activities often occur in different regions, their environmental impacts spill across both provincial and national boundaries. These spatial spillovers of burdens involve a degree of uncertainty because the precise locations of material extraction, manufacturing, operation, end-of-life treatment and other relevant life cycle processes are often unknown or represented by coarse regional averages, which potentially limits the accuracy of spatial environmental impact attribution. Our findings demonstrated that integrating more detailed supply chain data into LCI compilation can effectively reduce such uncertainty and improve the robustness of the results.
This study advances the application of spatially explicit LCA by developing and successfully demonstrating a practical and transferrable methodology framework. It also enriches existing LCI databases with comprehensive LCI datasets for China’s current and future power systems. The findings underscore the need to adopt dynamic mitigation strategies across both short- and long-term timeframes to achieve an overall sustainable transition. The spatial distribution and hotspot identification provide critical insights for more targeted, equitable and internationally coordinated mitigation strategies via both policymaking and supply chain management. By emphasising the value of higher spatial detail in LCIs, this work calls for prioritising improved data availability and transparent data-sharing among all stakeholders to encourage wider adoption of spatially explicit LCA in sustainable decision making.</p