Sydney eScholarship
Not a member yet
31878 research outputs found
Sort by
Capturing and understanding beef cattle production diversity in extensive environments to optimise productivity and welfare
No full textBeef production systems in Australia span diverse agro-climatic zones that are highly susceptible to
climate variability. Extreme climate events, particularly heatwaves and rainfall variability, can
adversely affect cattle production systems. Understanding how these events impact cattle
productivity across landscapes is, therefore, critical to support farmer decision-making. This thesis
advances understanding of the impacts of extreme climate events on cattle systems to help mitigate
their effects and optimise productivity and welfare. Chapter 2 presents a literature review highlighting
the impact of climatic extremes on cattle systems, advances in heat stress (HS) monitoring, feedbase
dynamics, and resilience frameworks to enhance productivity. Developing such frameworks is
essential to address the growing challenges posed by frequent and severe climate extremes.
Chapter 3 validates the Optiweigh™ (OW) in-field weighing station for remotely monitoring liveweight
(LW) in grazing conditions, providing a simple and accurate way to track LW changes through
voluntary animal attendance. Chapter 4 evaluates the temporal effects of HS on cattle systems,
defines thermal index thresholds, and explores adaptive strategies to sustain productivity under heat
stress. Insights from relatively homogenous indoor-housed dairy systems may serve as useful
benchmarks for managing HS in extensive beef systems. Chapter 5 assesses the impacts of
seasonal rainfall across climatic zones, revealing interactions between climate and feedbases,
contributing to improved management strategies. The delayed effects of HS and seasonal rainfall
(Chapters 4 and 5) underscore the need for adaptive strategies to mitigate climatic pressures and
sustain productivity and profitability. Overall, the findings demonstrate that quantifying the impacts of
extreme climate events enables adaptive, data-driven strategies to enhance the productivity and
welfare of extensive beef production systems
Hospital pharmacists, Aboriginal and Torres Strait Islander Peoples and chronic disease: helping to close the gap
No full textAboriginal and Torres Strait Islander Peoples are resilient, have survived and have actively prospered and cared for their communities for millennia. However, as a result of the ongoing effects of colonisation, Aboriginal and Torres Strait Islander Peoples are disproportionately affected by chronic diseases in Australia.
This thesis focussed on diabetes because for Aboriginal and Torres Strait Islander Peoples, diabetes is (1) often experienced from a young age; (2) associated with stigma and shame; (3) screened for using a readily accessible test (HbA1c); (4) has devastating complications which can be prevented by early screening and treatment options.
Diabetes like other chronic diseases causes preventable admissions to hospital for Aboriginal and Torres Strait Islander Peoples. It follows that hospital pharmacists have an opportunity to contribute to improve health outcomes for and together with Aboriginal and Torres Strait Islander Peoples. Review of the literature and a survey of hospital pharmacy departments nationally, did not illustrate extensive work in this area.
By working together with and listening to the priorities of Aboriginal and Torres Strait Islander Peoples, a new model of care was developed, implemented and evaluated. This was a novel pharmacist-led service that offered Aboriginal and Torres Strait Islander Peoples diabetes risk assessment and referral for specialist review, during their hospital stay. This service could help address identified missed opportunities to provide holistic hospital care.
This thesis produced a body of work to illustrate the importance of taking time to work together with Aboriginal and Torres Strait Islander Peoples to improve health outcomes. In doing this, factors required for wider application of such culturally safe models and future projects were identified. This thesis illustrated that success can only be achieved by ongoing collaborative partnerships together with Aboriginal and Torres Strait Islander Peoples
Using patient-reported outcomes to improve prognostication in advanced gastro-oesophageal cancer
No full textAdvanced gastro-oesophageal cancer is associated with poor prognosis and short, but variable, survival times. The goal of this PhD research was to improve prognostication for individuals with advanced gastro-oesophageal cancer by combining standard clinicopathological information from clinical trials with patient-reported outcomes (PROs).
We identified, organised, and summarised survival data from 44 randomised clinical trials, to provide estimated ranges for best-case, typical, and worst-case scenarios for survival time according to lines and types of treatment. This approach provides clinicians with information that they can use to estimate and explain scenarios for survival time to their patients seeking quantitative information about their prognosis.
In a separate scoping review, we identified, organised, and summarised 7 studies assessing the prognostic value of patient-reported outcomes as predictors of subsequent survival time in advanced gastro-oesophageal cancer. We found that appetite loss, pain, physical functioning, role functioning, social functioning, and global quality of life provided useful prognostic information in this setting.
We then developed and validated a multivariable prognostic model incorporating both PROs and standard clinicopathological features. We developed the model using data from INTEGRATE IIa (n=251) trial and validated it using INTEGRATE (n=152) data, demonstrating their predictive accuracy in an independent cohort.
This body of work provides coherent, structured strategies to help oncologists estimate and explain survival time and probabilities to patients with advanced gastro-oesophageal cancer seeking information about their prognosis. This research facilitates a personalised, patient-centred approach to prognostication that should support better-informed shared decision making, communication, survivorship care planning and overall patient care
Metal Extrusion Assays as a Tool for Sulfate Recognition
No full textSulfate is a biologically relevant anion found in an array of aqueous media including biological and
environmental fluids, including blood plasma and seawater. While sulfate is highly prevalent in these
media, selective detection of sulfate poses a challenge due to its similarities with other relevant
anions. While receptors specifically designed for sulfate have often been used, assay based
approaches can also be used for sulfate recognition. In this study, Metal Extrusion Assays (MEAs)
are used as a new approach for sulfate recognition. While this approach has been used for
recognition of other relevant anions, this study presents one of the first MEAs for sulfate in a range of
aqueous media.
To determine if this approach can be used for sulfate recognition in aqueous complex media, a range
of synthetic and analytical techniques were employed, with fluorescence and UV-vis spectroscopy used as the main tools to detect binding. The photophysical characteristics of a range of receptors
was determined in conjunction with studying these receptors as tools for cation binding. This then
informed the creation of the MEAs, which were employed to detect sulfate. Variation of the receptor-ligand
combination allowed for sulfate detection through MEAs in biological fluid mimics and seawater
mimics, indicating the potential of MEAs for use in complex media
Domain-Specific Cognitive Impairments Following Oxaliplatin and 5-Fluorouracil Treatment in Rats: A Preclinical Study
No full textThe FOLFOX chemotherapy regimen, combining oxaliplatin (OXA), 5-fluorouracil (5-FU), and leucovorin (folinic acid), is a global standard of care for colon cancer. However, both clinical and preclinical evidence of the cognitive side effects associated with this treatment remains largely mixed. Therefore, this study aimed to investigate the direct effects of combined OXA and 5-FU treatment on executive function in female Sprague Dawley rats, comparing different dosing protocols. In Experiment 1, rats received three weekly doses of OXA (6 mg/kg, i.p.) and were tested on a delayed non-matching-to-sample (DNMTS) task and the novel object recognition (NOR) tasks. OXA did not impair working memory but significantly impaired short-term recognition memory. Experiments 2 and 3 examined the effects of combined OXA and 5-FU treatment using different dosing regimens. Rats treated with two weekly doses of OXA (6 mg/kg, i.p.) and 5-FU (50 mg/kg, i.p.) exhibited significant deficits in set-shifting and spatial working memory but no impairments in reversal learning or recognition memory. Alternative regimens, including a single high-dose group (OXA 8 mg/kg + 5-FU 75 mg/kg) and a low-dose repeated group (OXA 6 mg/kg + 5-FU 50 mg/kg, spaced two weeks apart), produced minimal impairments, although the single-high dose group showed increased perseverative errors during set-shifting. These findings demonstrate that OXA and FOLFOX produce selective, domain-specific cognitive impairments in rats, with set-shifting and working memory particularly vulnerable to combined OXA and 5-FU treatment. These results highlight the need for targeted interventions to mitigate executive dysfunction in colorectal cancer survivors.
Raw data for this project.The dataset includes the summary statistics generated by the experiments reported in the manuscript 'Domain-Specific Cognitive Impairments Following Oxaliplatin and 5-Fluorouracil Treatment in Rats: A Preclinical Study'.
It describes the results of tests of executive function in laboratory rats, including delayed non-matching to sample (DNTMS) tests, novel object recognition (NOR) tests, spontaneous alternation in Y-maze (Y-maze) tests, and attentional set-shifting (AST) tests.
All procedures and described in the manuscript.
All protocols used in generating this dataset were approved by the University of Sydney Animal Research Ethics Committee, 2023/AE002378
THROAT study: tertiary hospital retrospective observational audit of tonsillectomy
No full textBackground
Adenotonsillectomy (AT) is the most common surgical treatment for paediatric obstructive sleep apnoea (OSA). Evolving patient demographics, diagnostic strategies and surgical techniques have prompted the need to reassess complication risks and outcome predictors in contemporary practice.
Aims
To evaluate trends in patient characteristics, surgical techniques and outcomes in paediatric AT over time; and to identify risk factors for post-operative complications and both parent-reported and PSG-confirmed OSA cure in order to develop a predictive tool for AT outcomes.
Methods
This retrospective cohort study reviewed 1,716 children undergoing AT at a tertiary paediatric hospital across two time-separated cohorts. Data were analysed for demographic trends, operative variables, post-operative complications and predictors of OSA resolution, using multivariable regression and subgroup analysis.
Results
The second cohort featured more complex patients, including higher rates of obesity, comorbidity and pre-operative OSA diagnoses. Partial tonsillectomy and coblation use increased, while cold dissection and oral antibiotic use declined. Despite increased patient complexity, general complication rates remained stable. Risk factors for complications included age <2, ASA ≥3, neuromuscular and syndromic conditions, developmental delay and local anaesthetic use. Adenoidectomy increased both complication risk and OSA cure likelihood. Reported oSDB cure rate remained stable while PSG-proven OSA cure rate increased by 38% over time with age 2-6, large tonsils and absence of comorbidities predicting better outcomes.
Conclusions
AT remains an effective treatment for paediatric OSA. Outcomes are influenced by both patient and operative factors. Risk stratification tools such as the proposed STARS model may improve clinical decision-making and resource allocation. These findings support a tailored approach to AT, balancing benefit and risk based on individual patient profiles
Label-Efficient Deep Learning with the Pre-trained Models
No full textThe test error of deep learning models often follows a power-law decrease with the increase of training data and model size. A powerful model requires collecting a significant amount of training data. Data collection, especially the labeled data, is expensive and labor-intensive. To address this issue, we propose to integrate active learning and semi-supervised learning on top of pre-training. Combining pre-trained models with active learning introduces challenges such as intensified phase transitions and increased training costs. The intensified phase transition implies that the types of samples to be selected change rapidly with the alteration of annotation quantity, so existing methods are effective only within a limited budget range of labeled data. To tackle this, we introduce a novel active learning strategy, Neural Tangent Kernel Clustering-Pseudo-Labels (NTKCPL), which allows active learning to directly estimate the model's empirical risk in the active learning pool. Additionally, based on the analysis of estimation errors, we propose a pseudo-label generation method to reduce the estimation error. Experimental results demonstrate that our approach outperforms existing methods and has a wider effective labeling budget range. Furthermore, since pre-trained models often large scale and require significant time to train, combining them with active learning significantly increases computational time. Therefore, we propose a new efficient active learning framework that improves active learning accuracy by aligning training methods and active learning features. Moreover, to further diminish the need for manual annotations, we introduce Active Self-Semi-Supervised Learning (AS3L). By comprehensively utilizing pre-training weight initialization, active sample selection, and semi-supervised learning guided by prior pseudo-labels, we greatly enhance the model performance when dealing with limited labeled data
Modeling Fine-grained Long-range Visual Dependency for Deep Learning-based Medical Image Analysis
No full textMedical image analysis has gained substantial attention among researchers and clinicians. The primary objective of medical image analysis is to exploit diagnostic and prognostic information from medical images to support clinical decision-making and personalized treatments, which serves as a broad systematic research topic covering a wide range of medical vision tasks, including both pixel-wise and image-wise medical image prediction tasks. Convolutional Neural Networks (CNNs) were widely used in early deep learning-based methods for their ability to extract hierarchical image features via translation-invariant convolution operations. Subsequently, transformers are becoming popular in medical image analysis, as they can capture long-range visual dependency in medical images via self-attention operations with global connectivity. Moreover, other network backbones, such as Multi-layer Perceptrons (MLPs) and State Space Models (SSMs), also emerged as alternatives to transformers in modeling long-range visual dependency in medical images. This evolvement of network backbones demonstrates the importance of modeling long-range visual dependency for medical image analysis. The objective of this thesis is to further investigate the modeling of long-range visual dependency for deep learning-based medical image analysis, specifically in identifying the gaps of existing methods in modeling long-range visual dependency and hence introducing new methods to advance medical image analysis via more effective long-range visual dependency modeling. First, this thesis presents the innovative use of transformers in both pixel-wise and image-wise medical image prediction tasks. Then, this thesis turns its focus to MLPs, where a novel MLP block is introduced to capture multi-range visual dependency. Finally, this thesis presents a comprehensive empirical investigation of MLPs in various pixel-wise medical image prediction tasks
Exploring Bacteriophage Therapy Against Pseudomonas aeruginosa in Paediatric Cystic Fibrosis: Early-Phase Clinical Trial Implementation
No full textThis thesis investigates bacteriophage therapy as a treatment for Pseudomonas aeruginosa infections in children with cystic fibrosis (CF). Despite significant advances in CF, P. aeruginosa remains an important pathogen, especially during pulmonary exacerbations, where it contributes to significant lung function decline. In a study of CF children with exacerbations, P. aeruginosa was isolated in a large proportion, with many not returning to baseline lung function and requiring readmission, despite antibiotic therapy. This highlights the ongoing challenge of managing chronic P. aeruginosa infections in CF.
The thesis also examines the process of isolating and purifying P. aeruginosa bacteriophages for therapeutic use. A nebulisation protocol was developed to preserve bacteriophage integrity during delivery, ensuring that bacteriophages remain viable for effective treatment. The first clinical trial of personalised bacteriophage therapy through bronchoscopic instillation followed by nebulisation in children and adolescents with CF is reported within, shows promising results, with no adverse effects and improvements in lung function. One participant achieved complete eradication of P. aeruginosa for the first time in six years, despite prior antibiotic treatments.
This research emphasises the need for innovative treatments to combat persistent P. aeruginosa infections and lays the groundwork for future clinical trials in bacteriophage therapy particularly in the battle againsts multi-drug resistant bacterial pathogens. The CHIP-CF study has expanded its criteria to include children as young as six, expanded indications to include all patients with suppurative lung disease, and plans to move to phase 2 trials, with international collaborations to further integrate bacteriophage therapy into clinical practice
The effect of anatomical structures on particle transport mechanisms in the human upper airway
No full textThe human upper airway exhibits complex and variable geometry, yet few studies have examined the impact of specific anatomical structures on drug deposition. This thesis systematically isolates key pharyngeal features—the uvula, epiglottis, and soft palate—to assess their influence on airflow and particle deposition in a realistic airway model.
Using Computational Fluid Dynamics (CFD), simulations of 3 µm particles revealed that widening the soft palate region significantly reduces upper airway deposition, while the removal of the uvula and epiglottis alters flow dynamics and regional deposition but has minimal effect on total pharyngeal deposition. These findings were validated through experimental studies using Next-Generation Impaction (NGI) and High-Performance Liquid Chromatography (HPLC), confirming that increased airway space enhances fine particle fraction (FPF) and reduces deposition.
Additionally, the study highlights that inhaler resistance and airway anatomy significantly impact lung dose efficiency. While NGI and HPLC effectively measure total deposition, they lack the capability for real-time regional deposition analysis. To address this, a novel laser-photodiode method was developed, enabling real-time tracking of aerosol dynamics, particle velocity, and regional deposition. This technique offers a practical, high-resolution approach for studying drug transport in the human airway, bridging the gap between experimental and computational models