70040 research outputs found
Sort by
Investigating novice translation students’ AI literacy in translation education
No full textRecent developments in AI translation have attracted renewed attention to the impact of technology in translation education and professional practice. However, specialised training in AI translation, including the latest generative AI and neural machine translation, is typically available only to postgraduate or upper-level undergraduate students. It may be naïve to believe that younger students are not already using AI translation tools prior to, and earlier in, their studies. Therefore, this study critically examined, via a questionnaire with open-ended questions, the AI literacy of novice students (n = 51) in the context of translation education, including their knowledge, usage, evaluation and ethical awareness. Our participants self-reported and exhibited knowledge pertaining to AI translation, with AI translation reported being utilised as a reference to understand the source text and provide suggested translations. Of significant concern is that some participants reported having more confidence in their competence to identify and correct errors produced by AI tools but were not aware of the possible negative impacts and ethical issues involved in AI. Building upon a critical discussion of our findings, we argue that AI translation and translation technology should be integrated from the onset of translation education to empower students to become more informed, critical and ethical professionals
Rationalising the inclusion of HDAC inhibitors with standard-of-care chemotherapy for high-risk neuroblastoma
No full textHigh-risk neuroblastoma is an aggressive, highly chemo-resistant childhood tumour. These patients will receive intensive, multi-modal therapy, although relapse with treatment resistant disease occurs in up to 50% of cases. We recently utilised mathematical modelling and longitudinal single-cell imaging to demonstrate that a non-genetic form of chemoresistance can arise in neuroblastoma through the impact of gene expression noise upon the stochastic nature of apoptotic signalling (Hastings*, Latham*, 2023, Science Advances). Within treatment naïve neuroblastomas, priming with the histone deacetylase (HDAC) inhibitor vorinostat could overcome this chemoresistance and sensitise tumours to treatment with specific standard-of-care chemotherapies. In order to further rationalise the inclusion of HDAC inhibitors with a wider range of standard-of-care chemotherapy treatments for high-risk neuroblastoma patients, we have now undertaken a functional analysis of a library of FDA-approved HDAC inhibitors. By using established high-content imaging assays and RNA sequencing (RNA-seq), this analysis has demonstrated that HDAC inhibitors with differing specificity are capable of eliciting a diverse range of cell behaviour in neuroblastoma tumours, which impacts the manner in which they should be deployed in a clinical setting. A key observation from this study was that the HDAC inhibitors belinostat and vorinostat did not directly induce apoptosis but readily primed the cells to allow for sensitisation to standard-of-care chemotherapies. These differing functional outcomes were also associated with unique histone acetylation patterns and mechanistically coherent transcriptional changes as determined by RNA-seq analysis.These mechanistic insights are now being leveraged to design rationalised treatment regimens that combine these HDAC inhibitors with standard-of-care chemotherapies. Optimal combinations were tested in two clinically relevant diagnosis patient-derived xenograft (PDX) models, identifying potential approaches capable of improving survival outcomes for high-risk neuroblastoma patients
An Edge-Aggregated Temporal Transformer with Dual Graph Neural Networks for Enhanced Dynamic Node Affinity Prediction
No full textDynamic node affinity prediction, which aims at forecasting the affinity values between a given node and all potential connections at a specific timestamp, poses significant challenges in dynamic graph analysis. These challenges arise not only from the large-scale nature of the datasets but also from the limitations of existing approaches, which are primarily tailored for dynamic link prediction. These methods usually merge edge features into a single temporal domain and ignore node-specific identities. Motivated by such observations, we propose EA DGNN, an edge-aggregated temporal transformer that integrates temporal encoding with dual graph neural networks. Our method is based on two core design principles. First, Edge-Aggregation for Temporal Consistency: by aggregating edge messages into regular snapshots, EA-DGNN preserves the unique identity features of each node and incorporates a temporal degree encoding scheme to capture the influence of node affinity over time. Second, Dual GNNs for Specialized Spatial Learning: by recognizing the heterogeneous nature of node interactions (e.g., between users and items), we employ dual graph neural networks that operate independently on different node types, thereby enabling tailored spatial feature extraction. Extensive evaluations on four benchmark datasets demonstrate that EA-DGNN outperforms not only state-of-the-art dynamic graph methods but also robust heuristic baselines, demonstrating its ability to effectively capture evolving node affinities under various temporal conditions
System Dynamics Land Combat Model
No full textThere are three primary gaps within the combat modelling literature where there is an under-representation: (1) medium-fidelity level modelling, (2) combat feedbacks, and (3) psychological states. The motivation of the research was to address these gaps and develop a generic, causal, dynamic, modern land combat model for the further understanding of dynamic tactical land warfare. The objective of the research is to build a medium-fidelity model of land combat incorporating current technologies.
The land combat model was built using the modelling methodology of system dynamics. The system dynamics modelling methodology follows the iterative best practise steps of: problem definition, hybrid diagramming, simulation modelling, model testing, and strategy war gaming.
The model is comprised from the following model sectors: (1) Direct Fires (kinetic effects) sector, (2) Combat Service Support (CSS) supply (sustainment and logistics), (3) Battlefield Control (spatial, position advantage, and manoeuvre) sector, (4) Situation Awareness (non-kinetic effects) sector, (5) Indirect Fires (kinetic effects) sector, and (6) Will-to-Fight (psychological states) sector. The six sectors are integrated together to form the System Dynamics Land Combat (SDLC) Model. All modelling assumptions are listed, with the model structure showcasing the combat feedbacks.
The model was built in collaboration with the Defence Science and Technology Group (DSTG). Confidence in the model was built through validating the Kherson (2022) conflict.
The contribution of the work satisfies the generic objective, as the model aligns with general Australian warfare concepts (find-fix-strike-exploit etc.) with broad application across scenarios spectrum and various current technology concepts. It can be rapidly adapted to different types of problems and scenarios. It is causal as it aligns with warfighting causal reasoning and what-if analysis. It satisfies the dynamic objective by incorporating warfighting time dimensions (feedback loops, events concurrency, and sequencing) into the model. It incorporates both qualitative and quantitative concepts, which allows for modern warfare complexity to be represented; qualitative concepts (morale, will-to-fight, intelligence etc.) as well as quantitative concepts (attrition rate, supply rate etc.) are included in the model. Lastly, the contribution of medium fidelity modelling allows for exploring current technologies evaluations and trade-off analysis for a broad range of new concepts across different parametrized scenarios
Design and Chacterisation of a Cough Simulator
No full textCoughs play a critical role in the transmission of infectious diseases, such as influenza and COVID-19, which spread through airborne particulate matter (PM). Recent studies have focused on understanding the flow dynamics of human coughs. However, investigating cough dynamics with human volunteers is challenging due to the inherent variability in coughs, even for a single subject, making consistent, repeatable measurements under specific ambient conditions difficult. This work aims to develop a tunable cough simulator to address this challenge.
A mathematical model was derived from a mass-spring-damper system to simulate cough dynamics. In this system, an extended spring is controlled by a motor and a rotational disk to pull bellows, generating a cough. The model successfully replicates specific cough profiles by determining the optimum disk size (d) and spring stiffness (k). These results highlight the novel tunability of the simulator design, allowing it to match specific human cough profiles.
Measurements from 12 subjects revealed a mean peak flow rate of 4.05 L/s. The results indicate that higher peak flow rates correspond to larger cough volumes. Tests on cough particulates showed that inhaling clean air (with near-zero particles in the 0.3-10 µm range) resulted in a 16.7% to 60.5% reduction in particle count compared to inhaling room air, demonstrating that inhaled particles influence exhaled particles during coughing.
A PM generator system was integrated into the simulator to replicate human cough PM distributions in the aerosolized particle size range. The simulator consistently matched the target cough PM size distribution, with an overall PM count of 11,423 ± 1,719 compared to the simulated PM count of 11,999 ± 984. Finally, the simulated cough flow field was measured using Particle Image Velocimetry to analyze the turbulent puff structure of the simulated cough
Key factors in women’s managerial advancement in the construction industry: insights from machine learning
No full textDespite ongoing efforts to promote gender diversity in the Australian construction industry, women remain significantly underrepresented in managerial positions. Differing from previous studies using traditional survey or interview approaches, this study applied career capital theory and analyzed 1,595 LinkedIn profiles with 11 features, related to work experience, network size, educational background, and industry recognition. Predictive modeling was conducted using MATLAB’s Classification Learner, applying multiple machine learning algorithms to assess the significance of those features in predicting managerial level. The results identified current employer size as the strongest predictor of female managerial levels. Women in small enterprises were more likely to reach top management, while those in large companies more likely remained in lower managerial levels. Experience duration also had a significant impact, but progression plateaued beyond seven years, indicating tenure alone does not drive advancement. Follower and connection count demonstrated a notable contribution, emphasizing the importance of professional visibility. Contrary to traditional assumptions, recommendation count and highest education level had lower relevance, while construction-related degrees, certifications, awards, and courses showed minimal impact. This study sheds light on the barriers and contributors of women’s managerial advancement and provides practical recommendations for policymakers and industry stakeholders to foster inclusive and equitable workplaces
Assessing the burden of severe nausea and vomiting of pregnancy or hyperemesis gravidarum and the associated use and experiences of medication treatments: An Australian consumer survey
No full textBackground There is little data on contemporary patterns of antiemetic use or women’s experiences when using such agents in the treatment of severe nausea and vomiting of pregnancy (NVP) or hyperemesis gravidarum (HG). Methods Online, national survey of Australian women who were currently or had previously experienced severe NVP or HG, distributed through the HG consumer group, Hyperemesis Australia between July and September 2020. Results There were a total of 289 respondents with a mean age of 33 years, of which 38% were currently pregnant. More than 50% of respondents reported “major impacts” of the condition on areas such as social life, ability to undertake daily chores, ability to eat or drink, effects on work, taking care of pre-existing children and sleep. This resulted in 62% of respondents reporting ‘often’ or ‘always’ experiencing feelings of depression or anxiety as a result of their HG symptoms, with 54% reporting considering terminating their pregnancy, and 90% having considered having no more children. The most commonly used anti-emetic was ondansetron (91%), followed by pyridoxine (62%), doxylamine (62%), and metoclopramide (61%). Nearly all (95%) women who reported using ondansetron commenced it within the first trimester, with 55% reporting use as a first-line therapy. Most women reported one or more side effects to anti-emetics such as headache, constipation, sedation or impaired cognition, with 31% stopping metoclopramide because of side effects, compared with 14% for ondansetron and 10% for doxylamine. Ondansetron, doxylamine and corticosteroids had the greatest perceived effectiveness, with more than 50% rating them as “effective” or “very effective”. Half (50%) reported use of acid suppressive therapy, with 51% reporting using complementary or alternative therapies in addition to conventional treatments. Conclusions The study findings demonstrate large variability in antiemetic use and outcomes, highlighting the need for individualised care and treatment approaches during pregnancy
ADVANCED ALGAL OXIDATION USING PLASMA BUBBLES: INSIGHT INTO HOW ALGAL CELL CHARACTERISTICS, DISCHARGE GAS PROPERTIES, AND SOLUTION PROPERTIES INFLUENCE THE OXIDATION MECHANISMS
No full textThe increasing occurrence of harmful algal blooms threatens water quality, necessitating innovative treatment solutions. Plasma bubbles, an advanced oxidation process, generate reactive species such as radicals, hydrogen peroxide (H2O2), and ozone (O3), offering a promising approach for algal oxidation. However, studies on how plasma bubbles and solution properties influence algal oxidation remain limited. This study investigates the underlying mechanisms driving algal oxidation in plasma bubble systems, focusing on the influence of algal cell characteristics, discharge gas composition, gas flow rates, and solution properties.
Experiments examined the response of green algae (Chlorella vulgaris) and cyanobacteria (Microcystis aeruginosa CS-555 and CS-564) to plasma bubbles. Flow cytometry showed low short-term removal efficiency (<5%) for all species, but cyanobacteria removal was 10–30% higher than green algae (<60%) over 72-168 h post-discharge, influenced by cell-specific degradation pathways. Scavenger tests revealed ∙OH radicals had limited impact on green algae, while cyanobacteria were highly susceptible to ∙OH radicals. Air-plasma bubbles were more effective for long-term green algae oxidation, whereas O₂-plasma bubbles enhanced cyanobacteria oxidation.
Gas flow rate studies using a venturi-plasma system showed that higher flow rates increased turbulence and superoxide generation, improving inactivation and removal. In contrast, lower flow rates preserved O3, enhancing long-term oxidation. Solution properties also influenced plasma bubbles efficacy. Higher cell concentrations initially enhanced reactive species interaction but limited long-term oxidation. Acidic conditions (pH 5–6) induced immediate oxidative stress. Increased salinity reduced oxidation efficiency due to reactive species quenching. These findings highlight the need to tailor plasma bubbles to algal species and treatment goals
Examining the Relationship Between Early-Life Stress, the Immune System, and the Potential for Intervention During Adolescence
No full textEarly life stress (ELS) has lasting effects on mental and physical health, potentially
through long-term changes to the immune system. Evidence suggests that ELS disrupts
neuro-immune development, leading to increased inflammation and greater vulnerability to
psychopathology. This thesis used a maternal separation (MS) model to examine the long-
term neuro-immune effects of ELS across of range of immune measures in adulthood.
Importantly, a second aim of this research was to examine whether an environmental
intervention during a later sensitive developmental period (specifically, adolescence) could
mitigate or reverse neuro-immune alterations associated with ELS. The experiments were
designed to build progressively upon one another, with the aim of building a comprehensive
understanding of the long-term immune consequences of MS and the potential for
intervention to mitigate these.
Chapter 2 examined the long-term effects of MS on anxiety-like behaviour, peripheral
pro-inflammatory cytokine expression, and microglial cells in adulthood, as well as the
potential for environmental enrichment (EE) to mitigate any MS-induced alterations. No
lasting effects of MS were observed on these measures, limiting the capacity to detect a
protective effect of EE.
Chapters 3 to 5 extended this investigation by providing a more comprehensive
assessment of neuro-immune functioning following MS. These experiments incorporated
additional immune markers, including an expanded panel of cytokines, more detailed
analyses of microglial morphology, and the inclusion of astrocyte measures. While no
consistent long-term alterations in cytokine expression were found, MS did result in enduring
changes to glial cell morphology, both microglia and astrocytes, with some preliminary
evidence suggesting that EE may partially ameliorate these long-term effects.
Chapter 6 concluded the experimental series by evaluating the long-term impact of
MS and EE on cytokine expression within the central nervous system. No significant effects
of MS or subsequent EE exposure were observed in brain cytokine expression.
Taken together, this thesis contributes to our understanding of how ELS impacts
neuro-immune function. Across experiments, there was some evidence of elevated
inflammation, increased sensitivity to subsequent stressors, and impaired regulatory
mechanisms. Importantly, EE showed some capacity to buffer against stress-related immune
reactivity, supporting its potential as a non-pharmacological intervention following ELS.
These findings are discussed in light of existing models linking early adversity and stress
sensitisation to long-term vulnerability for mental and physical health conditions
Exercise and Sports Science Australia updated position statement on exercise for preventing falls in older people living in the community
No full textBackgrounds: Falls affect a significant number of older Australians and present a major challenge to health care providers and health systems with over 380 older Australians hospitalised for a fall each day. Objectives: This statement seeks to inform and guide exercise practitioners and health professionals in safe and effective prescription of exercise to prevent falls amongst community-dwelling older people. Exercise prescription to prevent falls: Exercise is crucial for preventing falls in older age. Research evidence has identified that programmes which include functional balance and muscle strength training are the most effective in preventing falls. It is also important for exercise to be progressively challenging, ongoing and of sufficient dose to maximise its benefits in reducing falls. Additional (non-exercise) interventions are necessary for people with complex medical conditions, recent hospitalisation and/or particular risk factors not improved by exercise. People at a higher risk of falls may need greater support to undertake safe and effective fall prevention exercise. Global guidelines for fall prevention and management recommend that all older adults should receive advice about exercise to prevent falls. Qualified exercise professionals are well placed to prescribe and supervise functional balance and muscle strength training to older people with varied functional abilities, including those with co-morbidities