Filters
Availability
Publication Year
Resource Type
Faculty
School/Institute
Publication Search Results
Sort
Results per page
1 - 4 of 4
-
(2021)ThesisDementia is a leading cause of disability affecting approximately 50 million people worldwide. Currently, in Australia, there is no optimum way of monitoring the incidence or prevalence of dementia at the population level. There are also many unanswered questions regarding crucial aspects of dementia care, such as whether the provision of home-based services can reduce the time spent in residential care. Routinely collected administrative data have the potential to fill these gaps. This thesis explores the use of linked administrative data for detecting and monitoring dementia in Australia, uses these data to understand the care pathways followed by people with dementia, and addresses policy-focused questions aimed at improving dementia care. It does so by presenting the results of four research studies using the 45 and Up Study, a cohort of 267,153, recruited in 2006-2009 in New South Wales, Australia. The 45 and Up baseline survey was linked to a range of administrative datasets including records of hospitalisations, emergency department visits, aged care assessments, and claims for pharmaceuticals, medical services, aged care services and deaths for the period 2006-2016. Key findings include: (i) measuring dementia incidence with multiple linked administrative datasets identifies almost 80% of expected dementia cases (92% for those aged 80-84 years) and produces age-specific incidence rates that mirror those based on clinical diagnosis; (ii) entering residential care is the norm among people with dementia, and home-based care may not be meeting their needs at end of life; (iii) high-level home care for people with dementia may reduce the subsequent time spent in residential care; and (iv) changing to a new general practitioner (GP) when entering residential care is related to increased polypharmacy and initiation of psychotropic medicines among people with dementia. These findings will inform on-going efforts to monitor dementia incidence and care in Australia. They also have major policy implications, including emphasising the pressing need in Australia for more high-level home care packages, and highlighting end-of-life dementia care as a priority for policy development and innovation in service delivery. The link between GP continuity and psychotropic prescribing highlights a new intervention point that could assist in the efforts to reduce psychotropic prescribing in residential aged care.
-
(2021)ThesisThe application of machine learning models to big data has become ubiquitous, however their successful translation into clinical practice is currently mostly limited to the field of imaging. Despite much interest and promise, there are many complex and interrelated barriers that exist in clinical settings, which must be addressed systematically in advance of wide-spread adoption of these technologies. There is limited evidence of comprehensive efforts to consider not only their raw performance metrics, but also their effective deployment, particularly in terms of the ways in which they are perceived, used and accepted by clinicians. The critical care outreach team at St Vincent’s Public Hospital want to automatically prioritise their workload by predicting in-patient deterioration risk, presented as a watch-list application. This work proposes that the proactive management of in-patients at risk of serious deterioration provides a comprehensive case-study in which to understand clinician readiness to adopt deep-learning technology due to the significant known limitations of existing manual processes. Herein is described the development of a proof of concept application uses as its input the subset of real-time clinical data available in the EMR. This data set has the noteworthy challenge of not including any electronically recorded vital signs data. Despite this, the system meets or exceeds similar benchmark models for predicting in-patient death and unplanned ICU admission, using a recurrent neural network architecture, extended with a novel data-augmentation strategy. This augmentation method has been re-implemented in the public MIMIC-III data set to confirm its generalisability. The method is notable for its applicability to discrete time-series data. Furthermore, it is rooted in knowledge of how data entry is performed within the clinical record and is therefore not restricted in applicability to a single clinical domain, instead having the potential for wide-ranging impact. The system was presented to likely end-users to understand their readiness to adopt it into their workflow, using the Technology Adoption Model. In addition to confirming feasibility of predicting risk from this limited data set, this study investigates clinician readiness to adopt artificial intelligence in the critical care setting. This is done with a two-pronged strategy, addressing technical and clinically-focused research questions in parallel.
-
(2021)ThesisThe central aim of this thesis is concerned with the elucidation of cause-effect relationships from observational data among variables or events in the context of personalised healthcare. Appropriate methodology for extracting such relationships has been developed and discussed within the scope of longitudinal survival analysis. The thesis rethinks two fundamental questions of causality: (1) What empirical evidence is required for legitimate inference of cause-effect relationships? (2) Given that we are willing to accept causal information about a phenomenon, how can we draw inferences from such information? These questions have been without satisfactory answers in the light of observational healthcare, in part because we have not had clear definitions for the causal effect given the ever complex data structure and clinical questions on the unobserved outcome of survival probability, and in part because we have not had effective mathematical tools for deriving individualised causal answers to these questions. In the last decade, owing partly to advances in machine learning models, causality has undergone a major transformation: from a concept shrouded in properly designed randomised experiments into a mathematical object with well-defined semantics which can potentially be identified from observational data. This development has significantly brought down the cost of personalised healthcare in both medication discovery and evaluation. This thesis provides a systematic account of this causal transformation, addressed primarily to epidemiologists, particularly the pharmacoepidemiologist. Following a description of the conceptual and mathematical languages used in causal inference, this thesis emphasizes the development of practical methods for elucidating potentially causal relationships from (longitudinal) observational data, estimating the effects of treatment, and providing recommendations of treatments based on observed scenarios. I have tried in this thesis to present machine learning tools that handle causal relationships side by side with statistical probability theory. The prerequisites are startlingly simple, the results are straightforward. No more than basic skills in probability theory and some familiarity with machine learning are needed for the reader to begin solving causal problems that are too complex for the unaided intellect. The sequence of discussion follows more or less the chronological order by which I have tackled these topics, thus recreating for the reader the sense of progress that accompanied these developments. Following the introductory chapter (Chapter 1), I start with the formal description and review of the mathematical language of causal inference, in particular, how one can go about discovering cause-effect relationships using observational data (Chapter 2). I then proceed to the question of the definition of treatment effect on time-to-event outcomes. In particular, a framework of heterogeneous treatment effect estimation in survival analysis is provided to estimate the effects of static binary treatment conditions on a range of time-to-event outcomes encountered in the healthcare context (Chapters 3 and 4). The framework is then discussed in Chapters 5 and 6 with a more complex data stream, first in the case of time-varying confounders and exposures and then in the contingency of stochastic treatment options, where I examine the concepts of survival treatment effect given longitudinal confounding as well as the contour of time-varying survival dose response function. Chapter 7 offers an application of treatment effect estimation models in a recommendation system, where I used extensive simulations and case studies to demonstrate the advantage of an explanatory model over a predictive model in providing optimal treatment recommendations to patients. I end this thesis with the pursuit of the viability of causal exploration in public health management in Chapter 8, where a case study aiming to answer the question of the effect of public health interventions during the recent SARS-COV-2 pandemic is presented. The work shows that when there is a lack of quality observational data, approaches based on the potential outcomes theory (including the projects presented in Chapters 3 to 7) may fail to make a valid inference and additional assumptions have to be made on the functional form of the causal process. Together, my thesis bridges studies in causal inference theory and machine learning methods in the context of observational healthcare. Significant progress has been made in the development of estimation frameworks for individualised treatment effects subject to longitudinal data with informative censoring. I aim to provide pragmatic insights on the applications and limitations of advanced machine learning models in relation to empirical evidence and draw a clear distinction between causal discovery and conventional predictive learning in the realm of public health and epidemiology.
-
(2021)ThesisChapter 1 - Introduction. This provides an overview of mental and behavioural disorders (MBD) definitions and current knowledge about MBD in pregnancy. Maternal MBD in pregnancy are often overlooked. Most studies of MBD in pregnancy focused on a single class of MBD. This thesis used linked data from NSW Perinatal Data Collection and the NSW Admitted Patient Data Collection to assess diagnosed MBD in NSW maternities between 2002 and 2006. Chapter 2 – Methods. Describes data linkage, MBD definitions and preliminary data processing. Chapter 3 – Admissions for MBD in pregnancy. A study to compare rates of MBD admissions in pregnancy relative to MBD admissions in a baseline period. Overall, admissions for MBD were lower in early pregnancy (RR 0.71) and higher in late pregnancy (RR 1.91). Drug disorder admissions were more than 3-fold higher in late pregnancy. Schizophrenia admissions increased from early pregnancy and alcohol admissions remained lower throughout pregnancy. Baseline MBD admissions rates were higher for multiparous than primiparous maternities. Chapter 4 – Admissions with MBD in pregnancy. MBD prevalence in pregnancy was 2.4% overall, 1.4% for drug/alcohol disorders (DA) and 1.2% mental disorders (MD). Pregnancy DA prevalence was the same, psychotic disorder prevalence was half, affective disorder a third and anxiety a tenth that of comparable disorders in women of reproductive age. Coexisting MBD ranged from 23.6% for anxiety to 91.5% for sedative disorders. Smokers and residents in outer regional or more remote locations were identified as maternity populations at high risk of MBD. Chapter 5 – Neonatal outcomes. Assessed relative risks of individual classes of MBD on perinatal mortality, preterm birth, small size at birth, neonatal morbidity, and admission to neonatal intensive care (NICU). Adverse outcomes were on average 3- 4-fold higher for MBD relative to no MBD. Effects were universally attenuated by adjustment for smoking and co-existing MBD. Independent effects of opiate and cannabis disorders remained for most adverse neonatal outcomes, but not for schizophrenia or bipolar disorder. Chapter 6 – Discussion and conclusions. This thesis demonstrates the value of linked population data; has added to the evidence for pregnancy as risk for MBD; provided the first comprehensive prevalence estimates of MBD in pregnancy for all maternities in NSW, including both high and low prevalence MBD; provided evidence to support findings elsewhere of an independent association of alcohol, cannabis, or opiate disorder with poor neonatal outcomes, but not for schizophrenia or bipolar disorder.