70040 research outputs found
Sort by
Co-design of the Australian Prompt Response Network for a public-health focused intersectoral approach to information sharing on emerging drugs of concern
No full textThe rapid emergence of new psychoactive substances (NPS) and other emerging drugs of concern presents a significant global public health challenge, necessitating agile and interconnected drug information systems to identify and communicate risks. In Australia, responses have traditionally been localized, lacking a nationally coordinated system to rapidly share information about emerging drug threats. The National Center for Clinical Research on Emerging Drugs (NCCRED) collaborated with jurisdictional networks, clinicians, scientists, policy-makers, and peer organizations to co-design and co-produce the national Prompt Response Network (“PRN”). This process identified key components necessary to create an effective health-focused national network that supports and enhances existing and emerging jurisdictional and specialist early warning networks. The co-creation process resulted in several outputs, including a formalized national PRN group, an online knowledge exchange platform, a national website for disseminating drug alerts, and identified needs for a national drug signal database and an anecdotal reporting system. The PRN is the first Australian national public-health-focused mechanism for information exchange on new and emerging drugs and drug trends of concern. It provides the means for timely and responsive sharing of localized data, better informing risk assessment and facilitating a coordinated approach to public health responses and local and national preparation for emerging risks. Achieving this required mobilizing diverse disciplinary and community stakeholders toward a unified and collaborative response to preventing drug related harms
Classification and surveillance of Campylobacter in the genomic era
Get PDFDisease burden and financial cost associated with antimicrobial resistant bacteria pose significant challenges to the global public health system in the 21st century. Accurate classification of bacterial pathogens is essential for diagnosis, treatment and epidemiological surveillance. The advent of whole genome sequencing (WGS) at the turn of the century has revolutionised bacterial classification methodologies. Over the past two decades, the steadily decreasing cost of next-generation sequencing (NGS) has led to an explosion of high-quality WGS data in the public domain. This thesis contributed to the ongoing bioinformatics revolution in the genomic era, where the rapid accumulation of WGS data required advanced analytic tools to harness their full potential. By utilising the genomes of hundreds of thousands of bacterial isolates, the studies in this thesis redefined the species boundaries in the genus Campylobacter and developed novel typing schemes to improve long- and short-term epidemiological surveillance of C. jejuni.
The genus Campylobacter consists of species of both major and emerging pathogens. The taxonomy of the Campylobacter has progressed alongside the technological advancements since the genus was first recognised as human pathogens in the 1970s. An accurate, stable and universal genomic species definition for the genus is essential for both surveillance of established pathogens and detection of novel species.
In Chapter 2, a new tool was developed named Campylobacter genomic species typer (CampyGStyper) for assigning genomic species to Campylobacter isolates. A total of 498 publicly available Campylobacter genomes was initially used to define the Campylobacter genus core genome (500 loci). A core genome phylogenetic tree was then constructed with 2193 publicly available diverse Campylobacter genomes to assess the population structure within the genus. The core genome phylogeny was compared to the average linkage clustering
xi
of pairwise average nucleotide identity (ANI) similarity across 8440 diverse Campylobacter genomes, representing 33 species and eight subspecies. Species delineation at ANI 94.2% was found to be most consistent with the core genome phylogeny. Based on this founding, 60 ANI genomic species was defined that are consistent with previous comparative genetic studies. The validity of species delineation based on ANI clustering was further confirmed by pairwise in silico DNA-DNA hybridization, which scored below 70% between medoid genomes of each ANI clusters. The CampyGStyper had a species assignment accuracy of 99.96%, which is far superior compared to non-phylogenetic-based tools such as Kraken2. CampyGStyper has been published on the journal mSystems and is freely available to the public to facilitate assignment of established pathogens and detection of novel species in the genus Campylobacter (
https://github.com/LanLab/CampyGStyper).
In Chapter 3, a new tool was developed named, multilevel HierCC typing, to facilitate both long-and short-term C. jejuni epidemiological analysis. First, a C. jejuni cgMLST scheme was developed using 346 complete genomes and 2587 diverse assembled genomes. Statistical analysis was performed using cgMLST allelic profiles of 63,102 C. jejuni genomes to determine nine HierCC levels to offer a range of typing resolutions for different epidemiological analyses. These levels constitute the multilevel HierCC typing scheme. To showcase the application of the scalable typing scheme, country and continent specific clusters from HC264 to HC19 and farm-animal associated clusters from HC264 to HC55 were identified. The utility of HC4 for investigating C. jejuni transmission from farm animals to humans. was also demonstrated. The subset of key HierCC levels identified in this chapter allowed natural and genetically discrete clusters to be described consistently. Therefore, the multilevel HierCC typing scheme has the potential to be incorporated into a continuous, comprehensive and global epidemiological surveillance program for C. jejuni.
xii
Building on the C. jejuni multilevel HierCC typing scheme designed in Chapter 3, Chapter 4 integrated antimicrobial resistance (AMR) prediction with the multilevel HierCC typing scheme to investigate the global trend of C. jejuni AMR. Genomic AMR prediction was performed for a large C. jejuni global WGS dataset (n = 63012). The resistance of C. jejuni isolates to 15 different antibiotics and antibiotic classes were analysed for the global, the UK, and the USA dataset at different time frames. Resistant HierCC clusters were identified in the USA surveillance datasets from 2016-2022. The temporal trends of these clusters allowed differentiation between persistent and newly emerged resistant clusters. A hierarchical network of all resistant clusters in the USA surveillance dataset was constructed, which allowed tracking of infection sources of resistant strains and identification of AMR acquisition to multiple antibiotics. Analysis of macrolide-induced 50S_L22 mutations in the USA dataset was also performed. An increasing prevalence of 50S_L22 mutations over 2016-2022 and HierCC clusters carrying 50S_L22 mutations were identified, suggesting continuing selection pressure from macrolide use in the farm animals.
In summary, the three studies in this thesis have developed efficient bioinformatics workflows to process large WGS datasets. Novel bioinformatics solutions were developed for standardised species delineation of the Campylobacter genus and systematic epidemiological surveillance of C. jejuni. CampyGStyper is available publicly to provide accurate genomic species assignment for both major and emerging pathogens in the genus Campylobacter. The multilevel HierCC typing scheme meets the need for scalable typing resolutions to monitor both long- and short-term epidemiology of the major pathogen C. jejuni and standardised surveillance for C. jejuni AMR. Findings and associated tools developed in this thesis contribute to the fundamental understanding of Campylobacter and provide novel bioinformatics tools to facilitate the prevention and control of C. jejuni
Trends in stroke incidence, death, and disability outcomes in a multi-ethnic population: Auckland regional community stroke studies (1981–2022)
Get PDFBackground: Reliable data on trends of stroke incidence and outcomes over time are necessary for assessing the effectiveness of public health and clinical strategies, and for allocating healthcare resources. We assessed the levels and trends in incidence, mortality, early case fatality and disability for stroke in a defined, ethnically mixed population over 40 years. Methods: To analyse data from five population-based stroke incidence studies in adult residents (age ≥15 years) of the Greater Auckland Region of New Zealand (NZ) (1.35 million) over 12-month calendar periods for 1981–1982, 1991–1992, 2002–2003, 2011–2012, and 2021–2022. Fatal and non-fatal, hospitalised and non-hospitalised stroke events (first-ever and recurrent) were identified through multiple overlapping sources using clinical World Health Organization (WHO) diagnostic criteria and neuroimaging to define three major pathological types of stroke: ischaemic stroke (IS), primary intracerebral haemorrhage (PICH), subarachnoid haemorrhage (SAH), and stroke of undetermined type (SUT). Crude and age-standardised annual incidence, mortality, 28-day case fatality and disability level, and 40-year trends were calculated by age, sex, and ethnicity assuming a Poisson distribution. For comparison of our findings, we carried out a pooled analysis of methodologically comparable population-based stroke epidemiology estimates in high-income countries over the last two decades. Findings: Overall, there were 7462 first-ever strokes (9917 events) over the 40-year period (4,682,012 person-years). From 1981–1982 to 2021–2022, age-standardised stroke incidence rates decreased from 156/100,000 (95% confidence interval [CI] 143; 170) to 124/100,000 (119; 130) and mortality rates from 98/100,000 (88; 110) to 28/100,000 (26; 31) in nearly all age, sex, and ethnic groups. Moreover, from 2002–2003 to 2021–2022, there was an increase in stroke incidence of 1.28% per year (95% CI 0.38–2.17) in people aged 15–54 years, with the mean age of people with stroke decreasing from 73.0 (SD ± 13.8) in 2002–2003 to 71.6 (SD ± 14.9) in 2011–2012 and 70.7 (SD ± 15.2) years in 2021–2022 (p for trend <0.0001). The risk of stroke in Māori and Pacific people in 2021–2022 was almost 1.5 and 2.0 times greater than that in NZ Europeans. Ethnic disparities in the risk of stroke and age of stroke onset remained stable over the study period. From 1981–1982 to 2021–2022, 28-day stroke case fatality declined from 33.1% to 12.1% (p < 0.0001). There was a trend towards reducing 28-day case-fatality (from 31.6% [95% CI 27.6; 35.7] in 1981–1982 to 11.4% [10.0; 12.7] in 2021–2022) and an increasing proportion of stroke survivors with good functional outcome at discharge/28-days post-stroke (increased from 45.7% (95% CI 41.3; 50.0) in 1981–1982 to 60.2% (58.1; 62.3) in 2021–2022). Interpretation: Stroke incidence, 1-year mortality and 28-day case-fatality and disability have decreased in Auckland, NZ over the last 4 decades. However, over the last decade (2011–2022) there was a stagnation in the decline in the age-standardised stroke incidence rates. The absolute numbers of people with strokes, and those who have died or remained disabled from stroke, have significantly increased from 1981 to 2022. Ethnic disparities in the risk and burden of stroke persist. Effective prevention strategies for stroke must remain a high priority. Funding: Health Research Council of New Zealand
Reducing Clients’ Influence in Property Valuation in Nigeria: An Exploration of the Blockchain Technology
Get PDFProperty valuation is crucial in guiding and informing stakeholders of their property asset worth. A valuer, therefore, requires a certain level of professionalism, independence, and expertise to execute the task of property valuation per defined standards. Since property valuation outcomes influence decisions such as portfolio value, performance bonus, and credit applications, clients sometimes interfere with the valuation process to protect self-interest. The interference aptly tagged clients’ influence or pressure by extant literature has far-reaching implications, including mistrust in the ability of a valuer and inaccurate property values. Consequently, previous studies have proposed valuer integrity, ethics, peer review, regulations and information transparency as measures to reduce the impact of the self-seeking behaviour of clients. Despite these suggested measures, developing and low-transparent property markets such as Kenya and Nigeria, to name a few, show evidence of the persistence of client influence. Empirical studies from Nigeria report challenges associated with poor property market and macroeconomic indicators and relatively high corruption. Some studies provide evidence that property valuers may succumb to clients’ influence and that the suggested measures to address the challenge are not proving effective. Hence, using Nigeria as a case study, this research proposes a solution based on blockchain technology to address the challenge. Blockchain was chosen because the extant literature deems it a supplementary and complementary tool for e-governance and delivering transparency due to its nature as a digital distributed ledger for storing information and introducing trust in transactions.
Five research objectives were proposed: firstly, to assess the perception, extent and dimensions of client influence in property valuation in Nigeria. Secondly, to investigate the strengths and weaknesses of the property valuation process in Nigeria. Thirdly, to investigate the barriers and prospects to adopting blockchain technology in property valuation in Nigeria. Fourthly, to develop a blockchain technology tool and simulate the property valuation process, and finally, to validate the effectiveness of the proposed blockchain-based application in reducing clients’ influence in property valuation in Nigeria. The objectives were directed towards addressing the broad question of if blockchain technology could reduce clients’ influence in property valuation.
This research drew from the clients’ influence model, opportunistic principal theory, social power theory and technology-organisation-environment framework and was guided by a pragmatic philosophical approach. Through surveys, semi-structured interviews and a focus group discussion, quantitative and qualitative data were obtained from 202 participants (185 for the quantitative segment and 17 for the qualitative segment), which include property valuers, bankers, and proptech experts, all practising in Nigeria. In addition, an artefact, an Android mobile application operating on the Hyperledger Fabric Network, was built to test out blockchain technology in property valuation. The data analyses applied in this research included descriptive statistics, thematic analysis, fuzzy-DEMATEL, correlation analysis, mean score ranking and Chi-Square test for Independence.
Overall, the findings indicate that clients’ influence may be unavoidable for the Nigerian property valuation industry because of the conflicting interests and perceptions of stakeholders – property valuers, clients (financial institutions), and loan-seeking customers, and the environment that interplay to fuel the problem. At the same time, out of the 14 considered clients’ influence factors, the type of company, perception the public has of the industry, size of the firm, relationship with the client, type of client, and regulatory framework were causal factors of prominence. Concomitantly, the valuation professional organisations, Nigerian Institution of Estate Surveyors and Valuers (NIESV) and Estate Surveyors and Valuers Registration Board of Nigeria (ESVARBON) instituted guidelines and practices as currently in operation are grossly ill-equipped to address the problem. There is a low awareness of blockchain among property professionals, and barriers to its adoption were the knowledge, technical know-how of blockchain and the cost of implementing such technology. At the same time, blockchain technology showed prospects by delivering transparency in the simulated valuation process, which stakeholders perceived could reduce clients’ influence in Nigeria. Nevertheless, some potential challenges unique to the Nigerian environment emerged.
This research contributes to knowledge by identifying interrelationships between clients’ influence factors, revealing key barriers and prospects for blockchain technology in property valuation, and proposing and testing a technology-based tool which can be deemed an alternative to the legal-based approach established in the extant literature. Accordingly, countries and property markets of similar nature battling with clients’ influence in their property valuation industry can draw parallels from this research to craft a bespoke approach to address the issue
A Systematic Approach to Prioritise Diagnostically Useful Findings for Inclusion in Electronic Health Records as Discrete Data to Improve Clinical Artificial Intelligence Tools and Genomic Research
Get PDFAims: The recent widespread use of electronic health records (EHRs) has opened the possibility for innumerable artificial intelligence (AI) tools to aid in genomics, phenomics, and other research, as well as disease prevention, diagnosis, and therapy. Unfortunately, much of the data contained in EHRs are not optimally structured for even the most sophisticated AI approaches. There are very few published efforts investigating methods for recording discrete data in EHRs that would not slow current clinical workflows or ways to prioritise patient characteristics worth recording. Here, we propose an approach to identify and prioritise findings (phenotypes) useful for differentiating diseases, with an initial focus on relatively common small B-cell lymphomas. Materials and methods: A website enabling crowd-sourced recording of diseases and phenotypes was developed. An expert committee in the field of B-cell lymphomas standardised phenotype terminology for use in digital resources, and select terms were included in the Human Phenotype Ontology (HPO). A total of 100 patient lymph node biopsy samples were evaluated, and phenotypes were recorded as discrete data. Bayesian networks (BNs) were developed based on these data, and their diagnostic accuracy and ability to prioritise these phenotypes for inclusion in EHRs were assessed. Results: Out of 146 phenotypes identified from the website as potentially useful for differentiating four different lymphomas from each other and from benign lymph nodes, 70–75 were included in BNs. The diagnostic accuracy of different naïve BNs was 96.3% for non–marginal zone lymphoma cases and 50% for marginal zone lymphoma cases when all of the included phenotypes were used and 93.8% for non–marginal zone lymphoma cases and 27.5% for marginal zone lymphoma cases when only 15 phenotypes were included in the BNs. Conclusion: This pilot provides a starting point for systematic improvement and a dataset for comparing related approaches
Low-cost sensor networks and interventions to improve awareness and reduce exposure to air pollution (IP4.02.04)
Get PDFThis report summarises outcomes of work on high-efficiency particulate air (HEPA) plain English guidance under the NESP Sustainable Communities and Waste Hub project (IP4.02.04)
Optimisation and Utilisation of Focal Irreversible Electroporation in the Treatment of Localised Prostate Cancer
Get PDFRadical whole-gland therapies for localised prostate cancer – either radical prostatectomy or radiotherapy – provide excellent oncological control, however they also have significant toxicity, particularly urinary, sexual and bowel dysfunction. Focal therapy is a novel treatment modality that ablates only the area of the prostate containing cancer whilst preserving the unaffected prostate, thereby reducing side effects. There are multiple different energies currently under investigation for use in focal therapy.
Irreversible electroporation (IRE) is a novel non-thermal energy source that has been shown to have promising oncological and functional outcomes in early studies. This thesis aims to build on the previous limited evidence base, by investigating the use of IRE firstly in specific anatomic locations, secondly as a salvage therapy and thirdly the outcomes for salvage surgery following IRE-failure.
First, we evaluated the use of focal IRE in the extreme apex of the prostate. This is an area in the prostate where thermal ablative modalities are contra-indicated due to the potential of damaging the adjacent sphincter. An analysis of 50 patients treated with apical IRE showed that this energy modality provided excellent oncological control, without increasing rates of incontinence or erectile dysfunction.
Secondly, we assessed the outcome for robot-assisted radical prostatectomy (RARP) as salvage therapy for recurrence after focal IRE. We found that salvage-RARP was safe and effective and that the initial therapy with focal IRE did not compromise the patients’ subsequent oncologic or functional outcomes.
Thirdly, we performed a multi-centre, prospective single-arm trial investigating the outcomes of focal IRE in localised radio-recurrent prostate cancer. Radio-recurrent disease is notoriously difficult to treat and salvage-IRE in this trial was found in 37 patients to be safe with comparable oncologic outcomes, a lower complication rate and better functional outcomes compared to radical surger
Patient-Specific Blood Flow in a Population Using Fluid Dynamics Modelling
Get PDFCoronary Artery Disease (CAD) remains a leading global cause of morbidity and mortality. Given its widespread prevalence, various risk assessment tools, such as the Framingham Risk Score, have been developed. However, current risk models fail to explain disease development in more than a quarter of patients, particularly within specific groups based on factors such as age and gender.
Blood flow patterns, shaped by individual coronary geometries, play a crucial role in disease development. Atherosclerosis preferentially occurs in complex vessel segments due to adverse haemodynamics, such as low Wall Shear Stress (WSS), which affects endothelial function at different disease stages. Thus, an improved understanding of the influence of coronary geometries and blood flow could enhance current risk assessments.
However, previous studies have relied on limited patient-specific data, with less consideration of multiple geometric factors and diverse patient groups, such as sex-specific analysis, leading to ambiguous or conflicting conclusions. Thus, this thesis aims to overcome these knowledge gaps to enhance understanding.
This project used Computational Fluid Dynamics (CFD) to analyse coronary blood flow and was conducted in three phases. Phase 1 involved simulations on idealised and modified bifurcation models to investigate the effects of various shape factors on haemodynamics. Phase 2 extended this understanding to patient-specific arterial trees, considering both healthy and stenotic vessels. Phase 3 utilised a large-scale clinical dataset, GeoCAD (comprising 387 patients), to assess stenosis development and explore key risk factors, including demographic and clinical factors.
The findings demonstrated that diameter, curvature and torsion are key contributors to disease development. Haemodynamic influences varied across different disease stages, with flow features like helicity exhibiting a double-edged effect, contrasting with the previously reported favourable impact. Gender differences in haemodynamic factors, including WSS-based metrics and helical flow, underscored the importance of considering distinct patient groups. Additionally, simplified coronary models proved insufficient to capture the complex coronary blood flow, emphasising the need for large-scale simulations using patient-specific arterial trees. By addressing key limitations in the field, this research enhances the understanding of coronary blood flow and its role in disease progression
Graph Unbound: Toward Efficient Explicit and Implicit Graph Learning
Get PDFOver the past few decades, graphs have been extensively used to model real-world relationships. Explicit graph-based learning has emerged as a powerful paradigm for complex data across domains like social networks and financial systems. As a form of unstructured data, graphs are also utilized to model structured data, such as tabular data. Implicit graph-based learning shows promise in these contexts.
This thesis explores both explicit graph-based learning for community search and implicit graph-based learning for data quality management.
Firstly, we investigate explicit graph-based learning in three areas: 1) unsupervised community search; 2) attributed community search; and 3) multilayer community search. Unsupervised community search identifies a query-dependent community where nodes are densely interconnected without labels. We propose TransZero, a pre-trained graph Transformer framework with an offline pre-training phase and an online search phase, effectively locating communities without labels.
Attributed community search focuses on finding communities that maintain structural cohesiveness and attribute homogeneity. We introduce ALICE, which extracts candidate subgraphs and searches for communities using ConNet.
Multilayer community search aims to find query-dependent communities across layers of a multilayer graph. We design EnMCS, an ensemble-based framework that includes HoloSearch for identifying potential communities and EMerge, an ExpectationMaximization (EM)-based method for synthesizing these communities into a consensus community.
Secondly, we explore implicit graph-based learning for missing data imputation, a critical task for data quality. Missing data imputation fills in unobserved elements in the data matrix, aiming to match the complete dataset. We propose UnIMP, a Unified IMPutation framework leveraging large language models and high-order message passing to improve imputation for mixed data types. UnIMP employs a cell-oriented hypergraph to model tabular data and utilizes BiHMP, a bidirectional high-order messagepassing network, to aggregate information for accurate imputation.
We also explore the inherent variability in both missing data and missing mechanisms, proposing an uncertainty-driven network for missing data imputation, termed NOMI. NOMI includes a retrieval module, a neural network Gaussian process imputator (NNGPI) and an uncertainty-based calibration module, operating iteratively to enhance imputation performance