70040 research outputs found
Sort by
Alkaline Sulphite Delignification and Wood Densification: An Integrated Approach for Decontaminating and Repurposing CCA-treated Timber
No full textChromated copper arsenate (CCA) is a popular water-borne timber preservative in Australia. The preservative inhibits the biological attack from insects and microorganisms, hence extending the service life of the timber. Unfortunately, the disposal and recycling of spent CCA-treated wood present significant environmental and economic challenges due to the toxic nature of the preservative within the wood. Current disposal methods, such as landfilling and incineration, raise concerns about groundwater contamination and air pollution, necessitating the development of safe alternatives. This research explored a combined chemical and mechanical treatment process to remediate and repurpose CCA-treated wood through an alkaline sulphite delignification method, followed by densification to enhance its mechanical performance.
While alkaline sulphite treatment is commonly used for lignin removal, its feasibility in decontaminating CCA elements has not been explored. This study examined the potential of the alkaline sulphite process to extract CCA components and improve wood repurposing potential simultaneously. The effects of hydroxide concentration, sulphite concentration, and wood-to-solution ratio on the extraction of CCA elements and lignin were examined under low-temperature conditions (90°C). The process achieved significant removal efficiencies, with arsenic exceeding 97% and chromium up to 53%. However, copper extraction remained limited, with a maximum removal efficiency of only 10%. Despite operating at a low temperature, a delignification efficiency of 18% was achieved. The analysis of variance (ANOVA) revealed that the sulphite concentration was the primary factor influencing delignification, while both hydroxide and sulphite concentrations significantly affected arsenic and chromium removal.
This study also assessed the feasibility of recycling CCA-treated wood into durable composite materials through the alkaline sulphite delignification-densification method. This phase was conducted with a fixed hydroxide and sulphite concentration and an optimised wood-to-solution ratio (1g:10ml) based on previous findings. CCA-treated samples underwent alkaline sulphite delignification at varying reaction times (4–12 hours), followed by hot-pressing densification. The efficacy of CCA decontamination, the mechanical properties of the resulting composite material, and its environmental compliance were evaluated. Results showed that a 12-hour delignification treatment achieved the highest removal efficiencies of chromium (63%), arsenic (97%), copper (14%), and lignin (37.5%). Overall, the delignification process improved the compressibility in delignified wood samples (~67%), resulting in increased density (1.03 – 1.07g/cm3) and mechanical strength. The optimal mechanical properties were observed in 8-hour treated samples, with tensile strength increasing by 316% (207 MPa), shear strength by 250% (9 MPa), and bending strength by 218% (193 MPa) compared to untreated wood. However, samples treated beyond 8 hours exhibited strength deterioration, attributed to the loss of holocellulose (combined fraction of cellulose and hemicellulose) with prolonged exposure to alkaline conditions. Additionally, TCLP and SPLP tests confirmed that the recycled product met environmental safety standards.
These findings demonstrate the feasibility of recycling CCA-treated wood waste into high-strength, environmentally safe structural materials using the alkaline sulphite delignification-densification approach. Future research on the utilisation and recovery of valuable components from the spent alkaline sulphite liquor should be carried out to promote a circular economy
Detection of Fatigue-Induced Muscle Compensation: Towards Home-Based Rehabilitation
No full textWith the rapid advancement of technology such as wearable sensors, VR, exoskeletons, and end-effectors, there is a trend of optimizing the current rehabilitation workflow from clinical rehabilitation to home-based rehabilitation for mild to moderate conditions. While it brings convenience and reduces time and travel costs for the patients, the patients cannot receive direct, hands-on supervision from the healthcare provider. Without real-time feedback, patients may experience muscle compensation. The phenomenon utilizes alternative movement patterns to perform tasks when the primary muscle is fatigued. Performing a compensated technique brings negative long-term consequences. For example, it may further injure other muscles, potentially cause postural misalignment, and reduce the training effect.
This thesis addresses identifying and detecting muscle compensation in a home-based rehabilitation setting, explicitly focusing on fatigue-induced compensation in the bicep curl exercise. The research involves three key tasks: 1) Identify the biomechanical differences between standard and fatigue-induced bicep curl and establish the key indicator for muscle compensation; 2) Design a wearable strain sensor-based shoulder patch to capture the differences in real-time settings; 3) Integrate an Inertial Measurement Unit (IMU) to improve the compensation onset detection through Long Short-Term Memory (LSTM) sensor fusion. The following paragraphs detail the methodology and results of each task.
First, the shoulder joint kinematics range of motion (ROM) and the muscle synergy model are identified in both standard bicep curls and fatigue bicep curls. A subject study of 12 people was conducted using a motion capture system synchronized with a series of eight surface Electromyography (sEMG) sensors to capture their joint kinematics ROM, and the sEMG signal from various shoulder muscles while they conducted standard bicep curls and fatigue bicep curls using a 5kg dumbbell. The collected signal was then processed, and muscle synergies were derived to analyse the two states' differences comprehensively. The result indicated a shift in the relative contributions of the forearm muscles to the shoulder muscles and a more pronounced movement on the shoulder joint during fatigue.
Using these results, a wearable strain sensor-based shoulder patch is developed to detect fatigue-induced compensation. A preliminary study was first conducted to identify the sensor characteristics and the optimal placement that can reflect the subject’s fatigue condition. Results indicated that placing the sensor on the Acromioclavicular (AC) joint and observing the signal’s amplitude and variability can effectively reflect fatigue. Building on this observation, two algorithms that can effectively detect the subject’s fatigue-induced compensation in post-hoc and real-time settings were developed. The developed shoulder patch and algorithms were tested on 13 subjects conducting bicep curls using a 5kg dumbbell. The performance of the algorithm was compared with a motion capture system and sEMG sensors. Results reveal that the proposed sensors and algorithms can effectively capture the fatigue onset. However, the results also indicate space for improvement in real-time detection.
To enhance the performance of the proposed strain sensor-based shoulder patch, it was then integrated with an IMU. The integration was motivated by validation results from the previous study, which demonstrated that joint kinematics are effective in detecting the onset of fatigue. Moreover, the IMU is a widely adopted sensor for kinematics-based fatigue monitoring due to its portability, affordability, and reliability. We first collected the joint kinematics and strain sensor signals from 13 subjects while conducting bicep curl exercises using a 5 kg dumbbell. Then, an amplitude-based feature extraction algorithm based on the domain knowledge of the strain sensor signal and shoulder elevation depression ROM during fatigue in the bicep curl exercise was developed. The amplitude feature extracted was used to train an LSTM model. Experimental results indicated that the fusion of both IMU and strain sensors using LSTM outperformed the results presented in the previous study. The conducted research fills the necessary technological gaps in the home-based rehabilitation field and establishes a critical foundation in this field. The insights gained can be further extended to a broader rehabilitation application, enhancing remote patient monitoring and potentially apply to workspace environments, monitoring workers' conditions
Salt Supply Chain Practices and Technical Considerations for Switching to Potassium-Enriched Salt Use in Australia: A Qualitative Study
No full textBackground: Switching to potassium-enriched salt in food manufacturing and preparation is expected to lower blood pressure and improve stroke and cardiovascular disease outcomes. However, it is not well understood how the Australian salt supply chain would support increased use of potassium-enriched salt. Objective: To explore food industry insights and perspectives on (1) how the salt supply chain operates in Australia and (2) technical considerations for switching to potassium-enriched salt in food manufacturing and preparation. Methods: Twelve interviews with food industry stakeholders (including food, salt, and potassium-salt manufacturers) were conducted between March and July 2023, transcribed, and thematically analysed using a template analysis method in NVivo. Results: The regular sodium chloride salt supply chain in Australia was described as well-established, relatively simple, and mostly local, whereas the potassium-enriched salt supply chain was less established, separate from the sodium chloride supply chain, and reliant on imported ingredients. Most stakeholders, except potassium-enriched salt manufacturers, perceived cost, taste, and functionality as barriers to increased uptake of potassium-enriched salt as a food ingredient. Potential enablers included strengthening the potassium-enriched salt supply chain in Australia to withstand international changes, research and development, and retail promotion of potassium-enriched salt. Conclusions: Opportunities exist to strengthen food industry support for switching to potassium-enriched salt in Australia. Addressing feasibility and acceptability concerns through coordinated support from other sectors, along with incentives for industry adoption, is required. Industry uptake of potassium-enriched salt use in packaged food is essential to realising the full potential of this promising public health strategy
Disaster Recovery Almanac vol 2
No full textAnnual publication edited by the HowWeSurvive team at UNSW asking contributors to answer 'What do you think is the most pressing issue in disaster recovery that we need to be paying attention to?
Examining the influence of genetic variation across gene expression, methylation, and apolipoproteins in older adults.
No full textResearch studying complex traits and disease has elucidated many genetic variants but often their functional impact isunknown. One method to advance our understanding of the functional consequences of genetic variants is to identifyand characterise whether they act as quantitative trait loci (QTLs) – a genetic variant associated with a quantitative trait, such as gene expression QTLs (eQTLs), protein QTLs (pQTLs), and methylation QTLs (meQTLs). Given ourpopulation is ageing there is a need to understand the role genetic variants play in ageing-related traits and disease.
A review of the existing literature indicates that few studies have sought to catalogue blood QTLs, specifically eQTLs, pQTLs and meQTLs, in older adults. It also underscored the necessity of assessing various types of QTLs to elucidate the complex interplay between genetics and molecular processes, laying a solid foundation for this thesis’s empirical work. Therefore, the overall objective of this thesis was to identify and characterise blood eQTLs, pQTLs and meQTLs, with a focus on older adults.
In this thesis, the first empirical study identified and replicated thousands of blood eQTLs in two independentsamples of older Australians. The next research study focused on identifying pQTLs for seven apolipoproteins in older adults. Apolipoproteins are an important family of lipid transporting proteins that play vital roles in health and disease. Large GWAS meta- analyses, incorporating several independent cohorts, including older Australians, were performed and found significant genetic variants associated with the seven apolipoproteins. The final empirical studyinvestigated the intersect between the eQTL and pQTL results and in addition identified meQTLs using methylation sequencing data from our own in-house independent cohorts. This integrated approach offered a comprehensive view of the function of these QTLs, with a small subset of QTLs, acting as eQTLs, pQTLs and meQTLs. This work underscores the importance of integrating genetic and epigenetic data to improve our understanding of the complex biological pathways involved in ageing.
In conclusion, this work contributes original knowledge to the research field, by identifying and cataloguing three different types of blood QTLs, using primarily independent samples of older adults. The findings demonstrate valuable insights into the functions of SNPs that could be used to inform future research in this field, to uncover causalgenetic variants and potentially assist with the prediction of those at risk for ageing-related decline and disease, and for the development of strategies to improve health in our ageing population
Impact on women's body satisfaction of exposure to postpartum imagery on social media
No full textBackground: Social networking sites may be a convenient, accessible and low-cost option for delivering health information at scale to postpartum women. However, social media use is associated with decreased body satisfaction and may contribute to psychological ill-health. Our study aimed to determine whether exposure to body-focused imagery, typical of imagery targeting postpartum women on Instagram, is associated with a reduction in state body satisfaction and state body appreciation. Secondly, we aimed to determine whether including postpartum-health-focused imagery, in conjunction with body-focused imagery, is associated with improving state body satisfaction/appreciation, compared with no postpartum health content. Methods: A single blinded quasi-experimental survey study, recruiting women who had given birth in the previous 2-years, asked participants about key demographic information, social media use and assessed thin-ideal internalization and media appearance pressures using validated tools. Participants were then exposed to either (1) 15 body-focused images of women with a thin-average level of adiposity; (2) as per (1) PLUS 5 postpartum-health-focused images; or (3) as per (1) PLUS 15 postpartum-health-focused images. State body satisfaction/appreciation were assessed before and after image exposure. Results: State body satisfaction/appreciation did not change from pre- to post-image exposure in any groups and measures were not different between groups at any time point. Discussion: Short-term exposure to body-focused imagery typical of Instagram content targeting postpartum women may not alter state body satisfaction or state body appreciation. Furthermore, incorporating postpartum-health-focused imagery did not alter results. Further research investigating whether an intervention providing health information to postpartum women via social media platforms improves health outcomes may be warranted
Agentic Intelligence in Financial Markets: A Multi-Agent System with Embedded LLMs for Bilateral Fixed Income Trading
No full textThis thesis explores the integration of agent-based models (ABMs) with contemporary artificial intelligence inference methods -- specifically, Large Language Models (LLMs) -- to develop what has recently become known as `Agentic AI', or otherwise, `Agentic Intelligence'. The incorporation of agent-based model methods with LLMs holds great promise while raising many questions. Modelling complex dynamic financial trading networks, such as bilateral bond markets, is a challenging problem. In this thesis, I develop ABMs and incorporate LLMs, applying both to this particular problem domain. I examine the potential of such integration, including the challenges of using LLMs for modelling bilaterally traded fixed income markets, where the combination of limited data and complex human decision-making creates unique challenges.
My research reveals three key findings. Firstly, agent-based modelling demonstrates that multi-agent systems can effectively simulate strategic behaviours and autonomous decision-making in partially observable bilateral bond markets. My results reveal thresholds in the state of bilateral market stability, and demonstrate that market functioning depends more on agent diversity and cost structures than on the number of participants. These results validate the ability of agent architectures to capture complex market feedback mechanisms while providing quantitative insights for market design.
Secondly, LLMs, as components of Agentic Intelligence, demonstrate significant limitations in strategic decision-making, particularly in generating random responses and handling binary trading choices, exhibiting systematic biases and ``aversion patterns'' during complex decisions. Through systematic testing of LLM models in financial decision-making contexts, I demonstrate significant variations in output distributions between model versions and model sub-versions. Analysis reveals that methods of LLM sampling can critically impacts output and results -- and I introduce nomenclature to describe both. I successfully integrate both in the TraderTalk model I present, demonstrating that LLMs can realistically simulate trading interactions without domain-specific training. Challenges remain in coordinating agent turn-taking and output interpretation; observed discrepancies between trading intentions and execution enhance the model's realism by mimicking human trader observed behaviours.
Thirdly, I produce a bespoke ABM with an integrated LLM, the TRIBE(LLM) Model. This model provides a demonstration of viability and empirical substantiation for Agentic AI within complex financial market simulations. I showcase integrations of LLMs with agent-based models which enhance market simulations through more sophisticated client behaviour modelling. While LLM integration enriches agent decision-making processes, it also reveals critical system sensitivities -- notably, even minor trade aversion encoded in LLM requests can trigger complete market cessation. The complexity of variation in LLM binary decision-making produces distributions that are not well replicated with probabilistic or deterministic sampling -- opening the way for LLMs to provide more ``human-like'' unpredictability than current ABM methods. This integration models a shift of market power dynamics toward clients and introduces stochastic human-like variability that can lead to systemic market cessation, revealing emergent properties not captured solely by traditional ABM approaches alone.
These findings advance the theoretical understanding of multi-agent system design, detail and discuss challenges of LLM integration -- along with suggested implementation remedies, and finally, provide practical insights for modelling complex financial markets. My research demonstrates that whilst the combination of these technologies offers promising capabilities, successful implementation requires careful consideration of their respective strengths and limitations, particularly in harnessing the potential of Agentic AI to enhance agent autonomy in modelling market dynamics
Somatic Horizontal Transfer and the Impact of Bacterial Mobile Elements in Human Genomes
No full textThe de-repression of transposable elements in mammalian genomes is widely recognised for its contribution to genome instability, inflammation, and ageing, yet is viewed as a cell-autonomous event. In contrast to mammalian cells, prokaryotes constantly exchange genetic material through transposable elements, crossing both cell and species barriers, contributing to rapid microbial evolution and diversity in complex communities such as the mammalian gut. Here, we proposed and sought to investigate the hypothesis that transposable elements released from prokaryotes in the microbiome or from pathogenic infections could readily cross the kingdom barrier, integrating into the somatic genome of their eukaryotic hosts.
We suggest that horizontal acquisition of prokaryotic transposable elements into mammalian genomes has been concealed and consequently under-reported due to inherent limitations in existing genomics analytical pipelines, and we propose novel strategies to uncover these events. Initially, our investigation of uncharacterised human proteins in the Protein Data Bank reveals an unexpected discovery – a distinct insertion of a bacterial IS4-like transposase within an mRNA of a human uterine cancer cell line. We therefore developed a tailored bioinformatic pipeline to unearth evidence of somatic horizontal transfer in human genomes, utilising data from the 1,000 genomes project and several human cancer datasets.
Beyond computational analyses, we attempted to establish an in vitro model to simulate horizontal transfer by engineering a mutant Salmonella strain with a fluorescent reporter for detecting such events. Our final goal in developing this system was to develop an in vitro co-culture / infection model with mammalian cells aimed at demonstrating a proof of principle for the ability of bacterial mobile elements to integrate into the human genome.
Collectively, this work provides evidence to support the occurrence of somatic horizontal transfer from bacterial mobile elements to the somatic human genome. These findings shed light on a previously underexplored aspect of infection dynamics, and genomic interactions between evolutionary distant organisms
The promotion of child restraint use interventions integrated into prenatal and postpartum education through the health system in China
No full textRoad traffic injuries are a major public health issue in China, being the second leading cause of injury-related death among children and adolescents over the past 30 years. Although child restraint systems (CRS) are effective in reducing fatalities, their usage is low indicating a need for sustainable and accessible intervention to promote their use. The health system, with its extensive reach through prenatal and postpartum education, is a potential platform for integrating CRS use interventions.
This thesis reports seven studies. In Study 1, systematic search methods were used to conduct a scoping review to map existing child road safety policies in China, identifying gaps compared to WHO recommendations, especially for CRS-related policies. In Study 2, data from the Global Burden of Disease (GBD) study were used to identify the burden of road injuries among children under 20 years old in China from 1990 to 2021, across China and by subregions. Study 3 reports a systematic literature review to identify the types of interventions that have successfully promoted CRS use through the health system. In Study 4, a participatory, behaviour-theory-driven approach was used to develop a comprehensive CRS use intervention package. This was informed by qualitative exploration of perceived barriers to CRS use and expectations of a CRS use intervention among expectant parents, parents, and stakeholders from key prenatal and postpartum education institutions, such as schools for pregnant women in hospitals, community health service centres, and daycares. The intervention package includes educational materials and interactive lectures to address knowledge gaps and behavioural barriers. Study 5 used the Consolidated Framework for Implementation Research (CFIR) framework to identify barriers and facilitators to implementing the new intervention package through the health system. Study 6 iteratively piloted and refined the intervention with small groups of parents (n=84), evaluating its impact on their willingness to purchase CRS and their knowledge of child safety. Study 7 developed a protocol for a six-month cluster randomized controlled trial to evaluate the intervention among expectant parents and parents of children aged 0–3 years across four provinces with a planned sample size of 1730 expectant parents or parents.
The results provide a foundation for sustainable policy formulation and implementation of CRS interventions in China. The newly develope intervention package has the potential to improve parents' willingness and knowledge and can feasibly be implemented across China. Future work should focus on scaling up the intervention and monitoring their long-term impact on child road safety