Search

This thesis presents an approach to predicting a 3-dimensional, 3-component velocity field of a fluid ow that possesses a homogeneous dimension. At the core of this approach is the technique of stochastic estimation, which is commonly used to combine a small number of instantaneous measurements with previously acquired statistical data, to produce a prediction of the flow over a large number of locations. In the proposed technique, particle image velocimetry (PIV) is used to provide measurements for the stochastic estimation procedure, and the statistical stationarity along the homogeneous dimension of the flow is exploited to extend the use of stochastic estimation to provide a full volumetric prediction. The first section concerns the prediction performance of stochastic estimation. It is shown how the traditional approach to stochastic estimation is equivalent to ordinary least squares (OLS) regression. The properties of OLS, previously unconsidered in stochastic estimation literature, are presented, and shown to have a number of practical uses in the design and implementation of stochastic estimation procedures. Several alternative approaches to flow prediction are selected for further study, and their performance is compared in a series of trials, based on data from a numerically simulated channel flow. The newly-introduced biased techniques are shown to outperform or equal the accuracy of the stochastic estimation techniques across the entire range of parameters under investigation. The second section introduces the proposed volumetric prediction technique. A proof of concept is obtained using volumetric data from the simulated channel flow, and the resulting predictions show excellent quantitative and qualitative agreement with the original data. The predicted vortex ring data compares favourably with previous theoretical and experimental studies, and visualisation of the volumetric data appears to show the existence of secondary vortical structures around the outside of the ring core, which have previously only been observed in numerical simulations.

The thesis reports on a two-year study which looked at decision-making in two Scottish health boards. The research was prompted by: (1) a dearth of knowledge concerning the dynamics of decision-making in the National Health Service (NHS) at local .level; and (2) a desire to understand the factors at local level which might help to explain the persistence of resource imbalances between different sectors of health care despite numerous attempts to remove them. The case study, which forms the core of the thesis, has two aims: (1) to describe decision-making in the two boards in the area of development fund allocations (ie growth monies) in order to discover the process by which priorities are set; and (2) to provide insights into the operation of the management structure introduced when the NHS was reorganised in 1974. Reform was largely about improving the rationality of decision-making and the case study aims to show what this involved in practice. Rational and incremental theories of decision-making are reviewed for their usefulness in understanding events, although no single theoretical perspective is favoured over others. The main study findings were: (1) that, contrary to much popular opinion, decision-making in the NHS is, to a significant degree, decentralised - both boards possessed a large measure of influence over the services they provided; (2) that, despite this discretion, the emphasis in allocation decisions was on policy maintenance rather than on policy change; (3) that part of the explanation for this resided in a number of constraints, both external and internal, operating on decision-makers; and (4) that a number of coping strategies were adopted by decision-makers in order to facilitate the task of allocating resources and to come to terms with the uncertainty inherent in the decision-making environment. The research also revealed a number of tension points arising from the management structure, many of which served to reinforce the obstacles to change through development fund allocations.

Insulin-like growth factor-I (IGF-I) is the only known potentially modifiable risk factor for prostate cancer. Intake of dietary protein, especially from dairy products, might also be associated with risk and with circulating IGF-I, but it is not clear if amino acids play a role in these relationships. Moreover, investigations of circulating concentrations of metabolites might reveal novel risk factors for prostate cancer. This thesis investigates plasma concentrations of amino acids and other metabolites in relation to protein intake, IGF-I, and prostate cancer risk using data from the European Prospective Investigation into Cancer and Nutrition (EPIC). To characterise plasma metabolite profile in men consuming markedly different amounts and types of animal products (meat-eaters, fish-eaters, vegetarians and vegans), cross-sectional analyses of 392 men in the EPIC-Oxford sub-cohort were conducted. Of 21 amino acids, six varied significantly by diet group, and the metabolite profile of vegans was different from those of other diet groups owing to lower concentrations of several glycerophospholipids and sphingolipids. In a case-control study nested within EPIC, with a mean follow-up time of seven years, the relationship of plasma metabolites with risk of prostate cancer overall, by time to diagnosis, by tumour characteristics, and with risk of prostate cancer death, was investigated. Data from 1,077 matched sets suggested that seven metabolites, from various classes, were associated with risk of prostate cancer overall (p < 0.05). After correction for multiple testing, 12 glycerophospholipids were inversely associated with risk of advanced prostate cancer (the strongest OR1SD = 0.54; 95%CI: 0.40-0.72). In multivariate analyses, including data from 1,593 matched sets, principal component analysis (PCA) and treelet transform (TT) were used to identify patterns in metabolite profile, five of which were associated with risk of more aggressive tumour sub-types (high grade, advanced and aggressive disease) and/or prostate cancer death. There was a ≈ 50% lower risk of advanced and high grade prostate cancer in men with metabolite profiles characterised by high glycerophospholipids and sphingolipids (for advanced ORTT, top vs bottom third = 0:48; 95%CI: 0:31-0:74), with similar results for high grade and PCA). To investigate if associations between protein intake and circulating IGF-I may be mediated by plasma amino acid concentrations, cross-sectional analyses of amino acid concentrations with protein intake and IGF-I concentrations were carried out in 1,697 and 1,142 control participants, respectively, from the nested case-control study. Dairy protein intake was positively associated with concentrations of branched-chain amino acids and several other essential amino acids, while plant protein intake was strongly associated only with histidine. Serum IGF-I was positively associated with arginine and inversely with ornithine and certain amino acid ratios. In conclusion, men with different dietary habits with respect to the consumption of protein types have different amino acid and metabolite profiles, and metabolite concentrations may be associated with risk of more high-risk prostate cancer sub-types (high grade, advanced and aggressive disease) and prostate cancer death. Further large-scale studies are needed to determine if metabolites play a role in aetiology or are markers of sub-clinical prostate cancer.