Uppsatsen ämnar utvärdera olika estimatorers egenskaper för att beräknamedelvärdesskattningar för delvis censurerad miljödata erhållen från Sveriges LantbruksUniversitet (SLU). Detta undersöks genom en simuleringsstudie i vilken data har simuleratsfrån en χ2 fördelning där censureringsgrad och frihetsgrader har varierats. Därefter har deerhållna medelvärdesskattningarna utvärderats med hjälp av estimatorns medelkvadratfel(MSE) samt bias. Vidare har medelvärdesberäkningar gjorts på erhållen data från SLU. Allasimuleringar har gjorts i R, för medelvärdesberäkningar har tilläggspaketet “Nondetects AndData Analysis for environmental data” (NADA) använts. De estimatorer som utvärderats ärMaximum Likelihood (MLE), Regression on order statistics (ROS), Kaplan-Meier samtsubstitution. Resultaten visar att MLE samt ROS ger konstant bättre MSE och bias när dataantas följa en lognormalfördelning oavsett grad av censurering samt skevhet på data än vidnormalfördelningsantagande. Vid hög skevhet på data ger substitution resultat nära det sannamedelvärdet oavsett grad av censurering. De mest väntevärdesriktiga estimatorerna finnesvara ROS under lognormalantagande samt substitution när data är skevt fördelad, för ickeskevdata rekommenderas ROS med lognormalantagande, oavsett censureringsgrad.

Stability is a major requirement to draw reliable conclusions when interpreting results from supervised statistical learning. In this paper, we present a general framework for assessing and comparing the stability of results, that can be used in real-world statistical learning applications or in benchmark studies. We use the framework to show that stability is a property of both the algorithm and the data-generating process. In particular, we demonstrate that unstable algorithms (such as recursive partitioning) can produce stable results when the functional form of the relationship between the predictors and the response matches the algorithm. Typical uses of the framework in practice would be to compare the stability of results generated by different candidate algorithms for a data set at hand or to assess the stability of algorithms in a benchmark study. Code to perform the stability analyses is provided in the form of an R-package. / Series: Research Report Series / Department of Statistics and Mathematics

Background The central mission of every medical school is education, however most teachers have not received formal teacher training. Hence there is an expectation that faculty development (FD) with planned programmes to prepare faculty members for their role in teaching will be integral to medical school life. The widespread investment in FD is predicated on the belief that it enhances the effectiveness of teaching but the evidence is limited. Aim To carry out an in-depth exploration of FD in UK medical schools and evaluate the effectiveness of FD on teaching. The realist framework with its principle of explanatory causation was chosen to find out what works for whom in FD in what context, and why. This was an innovative use of the model in educational research. Methods A detailed literature review was carried out and combined with the researchers experience and insight of FD to develop eight realist hypotheses in the form of contexts, mechanisms and outcomes (CMO). Data that would support, modify or challenge the hypotheses were then collected in three phases. Phase I was a review and scoring of data on medical school FD webpages using a webpage scoring index. Phase II was observation and informal interviews of two cohorts of educators attending a FD course followed by detailed interviews of 12 educators (six from each cohort) six months later. Phase III was interview of FD coordinators and educators at eight medical schools. Quantitative and qualitative data were collected. Data analysis Quantitative data were analysed by descriptive methods, slope diagrams, bi-axial constructs and statistical analysis with Fisher’s exact test. Qualitative data were analysed by categorising and connecting strategies followed by summarising the relevant data under each hypothesis to check if the hypothesis was supported, modified or refuted. Findings The literature review revealed a paucity of publications on FD in the UK, however the 30 medical schools webpages reviewed showed data on FD activities. Data from all three phases were used to identify and explain the contexts (participatory approach/reflective practice, needs identification, supportive setting and standardization of medical teaching) that facilitated FD mechanisms (engagement, motivation, positive perception and professionalization).These led to the outcomes of improved confidence, competence, credibility and career progression. Conclusion Four realist theories of FD were identified (engagement, motivation, positive perception and professionalization) confirming the effectiveness of FD both in the short and long term and important recommendations for all FD stakeholders (educators, FD developers, universities and policy makers). In addition, I also made recommendations for future medical education researchers who choose the realist evaluation model.

Computational infrastructure for reprenting persons and citations has been available in R for several years, but has been restructured through enhanced classes "person" and "bibentry" in recent versions of R. The new features include support for the specification of the roles of package authors (e.g.,maintainer, author, contributor, translator, etc.) and more flexible formatting/printing tools among various other improvements. Here, we introduce the new classes and their methods and indicate how this functionality is employed in themanagement of R packages. Specifically, we show how the authors of R packages can be specified along with their roles in package 'DESCRIPTION' and/or 'CITATION' files and the citations produced from it. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Mining frequent itemsets and association rules is a popular and well researched approach for discovering interesting relationships between variables in large databases. The R package arules presented in this paper provides a basic infrastructure for creating and manipulating input data sets and for analyzing the resulting itemsets and rules. The package also includes interfaces to two fast mining algorithms, the popular C implementations of Apriori and Eclat by Christian Borgelt. These algorithms can be used to mine frequent itemsets, maximal frequent itemsets, closed frequent itemsets and association rules. (authors' abstract)

Particularism challenges the accepted idea of normative moral theory that morality can be reduced to a finite set of fundamental principles; it sees morality as quite capable of getting on without such principles. This thesis is concerned with asking what, if any, changes would be required in the practice of medical ethics if this is correct. It is proposed that current guidelines for professional clinicians and medical scientists constitute a “fleshed out” normative system which provides pro tanto rules for ethical practice. To investigate the implications of this in a particularist world, the idea of thin and thick moral concepts is extended to cover moral principles so that generalist professional guidance is seen as constituted of thick principles. This guidance aims to provide the required confidence for the doctor-patient relationship and in particular for the trust required between doctor and patient. Examples of the development of protocols for early phase clinical trials in cancer, and of resource allocation in a resource limited system are used to investigate the difference in decision making, and thus in the decisions themselves, between generalist and particularist professionals. In a generalist world trust is placed in the systems of trustworthiness (practice guidelines etc) and thus in the developers of such systems; in a particularist world moral decisions are made by the clinician and so trust is placed much more directly in that clinician. The implications of this analysis are that under particularism medical ethical training (initial and continuing) would focus more on the development of moral character of the various professionals and less of following guidelines. The complexity of modern medicine implies that such guidelines would still be required, but they would no longer represent pro tanto duties, but rather ceteris paribus advice.

Throughout our life cycle, the human body is exposed to harmful microorganisms. The innate immune system is a fundamental factor in the human body, which helps eliminate foreign organisms through specific signalling pathways with the involvement of immune receptors and signalling molecules. Viruses have evolved to infect the host and bypass the host immune responses,however a plethora of Pattern recognition receptors exist in the cell that are capable of detecting viral pathogens and mounting an innate immune response. Herpes simplex Virus Type 2 and Cytomegalovirus are common human pathogens that cause genital ulcerations, organ failure and mental health problems like encephalitis. In this study, we have aimed to identify the host’s innate immune response to HSV2 and HCMV infection in primary vaginal cells as well as Hela cells. Our data have shown that these viruses are recognized by TLR2 on the cell surface followed by intracellular PRRs such as TLR9, DAI, and IFI16, which trigger cytokine activation and release. Confocal imaging has revealed that these PRRs are located in different cell compartments and during viral cell entry and replication they can identify viral presence at specific parts of the cell. Therefore it seems that different PRRs are strategically placed in different cell locations to detect virus invasion and replication in order to activate cytokine secretion and protect the host. When different agonists for PRRs were used it was revealed that they were effective against Herpes virus infection thus indicating that a combination of PRRs agonists especially ones that trigger different cytokines could provide a wider spectrum prophylaxis to the host and they can be used to generate efficient treatment against HSV2 and HCMV infection.

A portfolio can be defined as “a purposeful combination of items” (Chien and Sainfort 1998). As the topic relates to research and development (R&D) the items in question are technologies, projects or products under consideration for inclusion in a given portfolio. As described by surveys from Cooper et al (1998), companies have widely varying practices for portfolio selection. This thesis examines existing literature to determine the key characteristics of good portfolio and portfolio method. The approach needs to handle multiple objectives, account for project interactions, and address the social aspect of decision making. The resulting portfolio should be aligned with business strategy, balanced, and of maximum value. It introduces general concepts that have been used to select single projects and reviews five specific applications and assesses them against the key characteristics from the literature. After identifying gaps in the current approaches, a comprehensive approach is proposed. This approach would (1) apply multi-attribute decision analysis at the portfolio level, (2) apply constraints for common inputs to cost such as resources, and (3) apply probabilistic methods to account for project interaction. This approach incorporates successful elements from existing approaches and addresses the two areas that are not adequately addressed with current approaches.

Stimulus control (SC) is a core component of cognitive behavioural therapy (CBT) for insomnia and is the single intervention for which there is most empirical evidence. Nonetheless, little is known about whether all of the elements within SC are critical to sleep improvement. This study, therefore, investigated the impact on sleep of the Quarter of an Hour Rule (QHR) a single, situational element considered central to SC for insomnia. The mechanisms of effect of SC intervention remain also unclear. An associated aim of the present study, therefore, was to contrast two forms of administration of the QHR to test aspects of the learning theory presumed to underlie the SC model. In addition adherence to the behavioural intervention was investigated and the possibility of using actigraphy to measure adherence objectively was explored. Prior to the randomised controlled trial (primary study), two preliminary studies were conducted. The first preliminary study aimed at determining the optimal cut-off to represent normalcy in sleep onset latency (SOL). The results indicated it to be fifteen minutes and, therefore, participants in studies two and three were asked to apply the QHR if they were not asleep within a quarter of an hour. Study two comprised three single cases and tested the feasibility of the QHR as a standalone therapy for insomnia. Visual inspection of the data and interrupted time series analyses evidenced SOL, wake after sleep onset (WASO) and sleep efficiency (S.E.) improvements in two out of three participants. Their Pittsburgh Sleep Quality Index (PSQI) score at the end of the intervention was reduced by 50% compared to baseline. The participant, whose sleep was not improved, following the intervention, had not applied the QHR. The results of this exploratory, single case, study warranted further investigation of the QHR. In study three forty-one GP and self referred volunteers, aged 18-72 years, with SOL and/or WASO complaints, formed 3 randomised groups: QHRin bed, QHRout of bed and control. Both QHR conditions required to ‘read if not asleep within a quarter of an hour’, with groups differing only with the location (in bed versus out of bed) where to apply the QHR. Sleep diary pre-treatment (two weeks) and post-treatment (three weeks), home polysomnography (PSG) (two nights pre-, two post-treatment) and sleep related questionnaire (pre and post) data were collected. Adherence with the QHR was measured objectively (actigraphy + light monitoring) and subjectively (adherence diary). Following QHR treatments, statistically significant reductions in SOL (QHRout) and WASO (QHRin and QHRout),an increase in S.E. (QHRin and QHRout) and a decrease in PSQI score (QHRin and QHRout) were found. Trends also indicated increased total sleep time (TST). Clinically significant improvements (SOL and WASO ≤ 31 minutes or reduced by 50%, PSQI ≤5 or reduced by 50%) were obtained in 33-57% of active groups participants.

Guidelines for the management of asthma in the UK have been published (BMJ, 1990, Thorax, 1993 and 1997) and embraced by many GP practices with improved outcome for patients. The study aims to observe and follow a cohort of adult asthma subjects from differing primary health care settings over a two-year period. Also to assess a newly devised patient focused morbidity index (Q score) by comparison to an established asthma specific quality of life questionnaire (AQLQ, Juniper et al, 1993). One hundred and fourteen subjects from four GP Practices, two inner city and two suburban were studied. Morbidity was assessed by AQLQ and Q score (Rimington et al, 2001), psychological status by the hospital anxiety and depression (HAD) scale (Zigmond and Snaith, 1983). Spirometry values (forced expiratory volume in one second, FEV1), peak expiratory flow (PEF) and details of current treatment as per BTS guidelines treatment step were recorded as markers of asthma severity. Subjects were assessed at baseline, twelve and twenty-four months. A random sub set of patients was asked to repeat certain elements of the study protocol at two weeks in order to assess the reliability of the Q score. The Q score correlated from baseline to two weeks (rs=0.61) as did AQLQ symptom score (rs=0.74) both p<0.01. At baseline AQLQ symptoms correlated with PEF (rs=0.40, p<0.001) and with BTS guidelines treatment step (rs=0.25, p=0.001) as did the Q score. Similar levels of correlation were reported for FEV1 with symptoms. HAD scores also correlated to AQLQ and Q score, but there was little correlation with lung function. At one and two year follow up no significant differences were observed in subjective or objective markers of asthma. There was a significant increase (p<0.001) in the number of subjects in the higher BTS guidelines treatment steps from baseline to twelve and twenty-four months while psychological symptoms remained high for inner city patients. In conclusion the Q score yields similar results to the AQLQ and is quick and easy to use in any busy clinic. The GP practice, at the forefront of asthma care should be offering appropriate therapy and regular review. The Q score used as a patient focused morbidity index can be a useful audit tool. Altering medication can give the impression of treating asthma but with out short-term reassessment the same levels of morbidity can persist.