Regression analysis of compositional data when both the dependent variable and independent variable are components

It is well known that regression analyses involving compositional data need special attention because the data are not of full rank. For a regression analysis where both the dependent and independent variable are components we propose a transformation of the components emphasizing their role as dependent and independent variables. A simple linear regression can be performed on the transformed components. The regression line can be depicted in a ternary diagram facilitating the interpretation of the analysis in terms of components. An exemple with time-budgets illustrates the method and the graphical features

Estimation of lung cancer diagnosis and treatment costs based on a patient-level analysis in Catalonia (Spain)

Background: Assessing of the costs of treating disease is necessary to demonstrate cost-effectiveness and to estimate the budget impact of new interventions and therapeutic innovations. However, there are few comprehensive studies on resource use and costs associated with lung cancer patients in clinical practice in Spain or internationally. The aim of this paper was to assess the hospital cost associated with lung cancer diagnosis and treatment by histology, type of cost and stage at diagnosis in the Spanish National Health Service. Methods: A retrospective, descriptive analysis on resource use and a direct medical cost analysis were performed. Resource utilisation data were collected by means of patient files from nine teaching hospitals. From a hospital budget impact perspective, the aggregate and mean costs per patient were calculated over the first three years following diagnosis or up to death. Both aggregate and mean costs per patient were analysed by histology, stage at diagnosis and cost type. Results: A total of 232 cases of lung cancer were analysed, of which 74.1% corresponded to non-small cell lung cancer (NSCLC) and 11.2% to small cell lung cancer (SCLC); 14.7% had no cytohistologic confirmation. The mean cost per patient in NSCLC ranged from 13,218 Euros in Stage III to 16,120 Euros in Stage II. The main cost components were chemotherapy (29.5%) and surgery (22.8%). Advanced disease stages were associated with a decrease in the relative weight of surgical and inpatient care costs but an increase in chemotherapy costs. In SCLC patients, the mean cost per patient was 15,418 Euros for limited disease and 12,482 Euros for extensive disease. The main cost components were chemotherapy (36.1%) and other inpatient costs (28.7%). In both groups, the Kruskall-Wallis test did not show statistically significant differences in mean cost per patient between stages. Conclusions: This study provides the costs of lung cancer treatment based on patient file reviews, with chemotherapy and surgery accounting for the major components of costs. This cost analysis is a baseline study that will provide a useful source of information for future studies on cost-effectiveness and on the budget impact of different therapeutic innovations in Spain.

MENTAL FILES AND METAFICTIVE UTTERANCES

ABSTRACT Metafictive utterances raise a kind of intuitions (intuitions of truthfulness) that pose a problem for a view that combines a referentialist approach to proper names with an antirealist stance on fictional characters. In this article I attempt to provide a solution to this problem within the framework of mental files. According to my position, metafictive utterances literally express an incomplete proposition while pragmatically conveying a complete one, which accounts for the intuitions of truthfulness. The proposition pragmatically conveyed is ‘metarepresentational', I'll argue, in the sense that it is about a mental representation or mental file.

Choosing the optimal energy system for buildings and districts

The purpose of this master’s thesis was to develop a method to be used in the selection of an optimal energy system for buildings and districts. The term optimal energy system was defined as the energy system which best fulfils the requirements of the stakeholder on whose preferences the energy systems are evaluated. The most influential stakeholder in the process of selecting an energy system was considered to be the district developer. The selection method consisted of several steps: Definition of the district, calculating the energy consumption of the district and buildings within the district, defining suitable energy system alternatives for the district, definition of the comparing criteria, calculating the parameters of the comparing criteria for each energy system alternative and finally using a multi-criteria decision method to rank the alternatives. For the purposes of the selection method, the factors affecting the energy consumption of buildings and districts and technologies enabling the use of renewable energy were reviewed. The key element of the selection method was a multi-criteria decision making method, PROMETHEE II. In order to compare the energy system alternatives with the developed method, the comparing criteria were defined in the study. The criteria included costs, environmental impacts and technological and technical characteristics of the energy systems. Each criterion was given an importance, based on a questionnaire which was sent for the steering groups of two district development projects. The selection method was applied in two case study analyses. The results indicate that the selection method provides a viable and easy way to provide the decision makers alternatives and recommendations regarding the selection of an energy system. Since the comparison is carried out by changing the alternatives into numeric form, the presented selection method was found to exclude any unjustified preferences over certain energy systems alternatives which would affect the selection.

Vaccination Competence - The Concept and Evaluation

The purpose of this two-phase study was to define the concept of vaccination competence and assess the vaccination competence of graduating public health nurse students (PHN students) and public health nurses (PHNs) in Finland, with the goal of promoting and maintaining vaccination competence and developing vaccination education. The first phase of the study included semi-structured interviews with vaccination professionals, graduating PHN students and clients (a total of n=40), asking them to describe vaccination competence as well as the factors strengthening and weakening it. The data were analyzed through content analysis. In the second phase of the study, structured instruments were developed, and vaccination competence of PHN students (n=129) in Finland and PHNs (n=405) was assessed using a self-assessment scale (VAS) and taking a knowledge test. PHNs were used as a reference group, enabling us to determine whether a satisfactory level of vaccination competence was achieved by the end of studies, or whether it was gained through work experience vaccinating clients. The data were collected from five polytechnic institutions and seven health centers located in various parts of the country. The data were collected using instruments developed for this study, and were analyzed statistically. In the first phase, based on the results of the interviews, vaccination competence was defined as a large multi-faceted entity, including the concepts of competent vaccinator, competent implementation of the vaccination, and the outcome of the implementation. Semi-structured interviews revealed that factors strengthening and weakening vaccination competence were connected to the vaccinator, the client being vaccinated, the vaccination environment and vaccinator education. On the whole, factors strengthening and weakening vaccination were the opposite of each other. In the second phase, on the self-assessment of vaccination competence, students rated themselves as significantly lower than working professionals. On the knowledge test, the percentage of correct answers was lower for students than PHNs. When all background variables were taken into account in multivariate analysis, there was no longer a significant difference between the students and PHNs on the self-assessment. However, in multivariate analysis, the PHNs still performed better than students on the knowledge test. For this study, a satisfactory level of vaccination competence was defined as a mean of 8.0 on the self-assessment and 80% correct answers on the knowledge test. Based on these criteria, students almost reached the level of satisfactory in their overall self-assessment, and PHNs did. Both groups, however, did rank themselves as satisfactory in some sum variables. On the knowledge test the students did not achieve a level of satisfactory (80%) in their total score, though PHNs did. As before, both groups did achieve a level of satisfactory in several sum variables. Further research and development should focus on vaccination education, the testing of vaccination competence and vaccination practices in clinical practice, as well as on developing the measurement tools.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

A bridge between engineering and language learning : automation and German in an online course

Artikkeli luettavissa osassa: Part 2. - ISBN 9789522163172(PDF). - Liitteenä työpaperi

Metabolism and gene polymorphisms of the folate pathway in Brazilian women with history of recurrent abortion

PURPOSE: To investigate the association between polymorphisms in genes that encode enzymes involved in folate- and vitamin B12-dependent homocysteine metabolism and recurrent spontaneous abortion (RSA).METHODS: We investigated the C677T and A1298C polymorphisms of the methylenetetrahydrofalate reductase gene (MTHFR), the A2756G polymorphism of the methionine synthase gene (MS) and the 844ins68 insertion of the cystathionine beta synthetase gene (CBS). The PCR technique followed by RFLP was used to assess the polymorphisms; the serum levels of homocysteine, vitamin B12 and folate were investigated by chemiluminescence. The EPI Info Software version 6.04 was used for statistical analysis. Parametric variables were compared by Student's t-test and nonparametric variables by the Wilcoxon rank sum test.RESULTS: The frequencies of gene polymorphisms in 89 women with a history of idiopathic recurrent miscarriage and 150 controls were 19.1 and 19.6% for the C677T, insertion, 20.8 and 26% for the A1298C insertion, 14.2 and 21.9% for the A2756G insertion, and 16.4 and 18% for the 844ins68 insertion, respectively. There were no significant differences between case and control groups in any of the gene polymorphisms investigated. However, the frequency of the 844ins68 insertion in the CBS gene was higher among women with a history of loss during the third trimester of pregnancy (p=0.003). Serum homocysteine, vitamin B12 and folate levels id not differ between the polymorphisms studied in the case and control groups. However, linear regression analysis showed a dependence of serum folate levels on the maintenance of tHcy levels.CONCLUSION: The investigated gene polymorphisms and serum homocysteine, vitamin B12 and folate levels were not associated with idiopathic recurrent miscarriage in the present study. Further investigations are needed in order to confirm the role of the CBS 844ins68 insertion in recurrent miscarriage.

Taxonomy, species richness and biogeography of Finnish crane flies (Diptera, Tipuloidea).

Biodiversity is unequally spread throughout terrestrial ecosystems. The highest species richness of animals and plants is encountered around the Equator, and naturalists observe a decrease in the number of creatures with increasing latitude. Some animal groups, however, display an anomalous species richness pattern, but these are exceptions to the general rule. Crane flies (Diptera, Tipuloidea) are small to large sized, non-biting nematoceran insects, being mainly associated with moist environments. The species richness of crane flies is highest in the tropics, but these insects are species rich and abundant in all biogeographic realms, boreal and arctic biomes included. The phylogeny and systematics of crane flies are still at an early stage and somewhat controversial. New species are constantly discovered even from temperate Europe, faunistically the best known continent. Crane flies have been rather neglected group of insects in Finland. The history of Finnish crane fly taxonomy and faunistics started in 1907, the year when Carl Lundström published his two first articles on tipuloids. Within roughly 100 years there have been only a handful of entomologists studying the Finnish fauna, and the species richness and natural history of these flies have remained poorly understood and mapped. The aim of this thesis is to clarify the taxonomy of Finnish crane flies, present an updated and annotated list of species and seek patterns in regional species richness and assemblage composition. Tipula stackelbergi Alexander has been revised (I). This species was elevated to a species rank from a subspecific rank under T. pruinosa Wiedemann and T. stackelbergi was also deleted from the list of European crane flies. Two new synonyms were found: T. subpruinosa Mannheims is a junior synonym of T. freyana Lackschewitz and T. usuriensis Alexander is a junior synonym of T. pruinosa. A new species Tipula recondita Pilipenko & Salmela has been described (II). Both morphology and COI (mtDNA) sequences were used in the assessment of the status of the species. The new species is highly disjunct, known from Finland and Russian Far East. A list of Finnish crane flies was presented, including the presence of species in the Finnish biogeographical provinces (III). A total of twenty-four species were formally reported for the first time from Finland and twenty-two previously reported species were deleted from the list. A short historical review on the studies of Finnish crane flies has been provided. The current list of Finnish species consists of 338 crane flies (IV, Appendix I). Species richness of all species and saproxylic/fungivorous species is negatively correlated with latitude, but mire-dwelling species show a reversed species richness gradient (i.e. an increase in the number of species toward north). Provincial assemblages displayed a strong latitudinal gradient and faunistic distance increased with increasing geographical distance apart of the provinces. Nearly half (48 %) of the Finnish crane flies are Trans-Palaearctic, roughly one-third (34 %) are West Palaearctic and only 16 and 2 % are Holarctic and Fennoscandian, respectively. Due to the legacy of Pleistocene glaciations, endemic Fennoscandian species are problematic and it is thus concluded that there are probably no true endemic crane flies in this region. Finally, there are probably species living within Finnish borders that have hitherto remained unnoticed. Based on subjective assessment, the number of “true” (i.e. recorded + unknown species) species count of Finnish crane flies is at minimum 350.

PET Imaging of Osteomyelitis. Feasibility of 18F-FDG, 68Ga-Chloride and 68Ga-DOTAVAP-P1 Tracers in Staphylococcal Bone Infections

Due to technical restrictions of the database system the title of the thesis does not show corretly on this page. Numbers in the title are in superscript. Please see the PDF-file for correct title. ---- Osteomyelitis is a progressive inflammatory disease of bone and bone marrow that results in bone destruction due to an infective microorganism, most frequently Staphylococcus aureus. Orthopaedic concern relates to the need for reconstructive and trauma-related surgical procedures in the fast grow¬ing population of fragile, aged patients, who have an increased susceptibility to surgical site infections. Depending on the type of osteomyelitis, infection may be acute or a slowly progressing, low-grade infection. Peri-implant infections lead to implant loosening. The emerging antibiotic resistance of com¬mon pathogens further complicates the situation. With current imaging methods, significant limitations exist in the diagnosing of osteomyelitis and implant-related infections. Positron emission tomography (PET) with a glucose analogue, 18F-fluoro¬deoxyglucose (18F-FDG), seems to facilitate a more accurate diagnosis of chronic osteomyelitis. The method is based on the increased glucose consumption of activated inflammatory cells. Unfortunately, 18F-FDG accumulates also in sterile inflammation regions and causes false-positive findings, for exam¬ple, due to post-operative healing processes. Therefore, there is a clinical need for new, more infection-specific tracers. In addition, it is still unknown why 18F-FDG PET imaging is less accurate in the detec¬tion of periprosthetic joint infections, most frequently due to Staphylococcus epidermidis. This doctoral thesis focused on testing novel PET tracers (68Ga-chloride and 68Ga-DOTAVAP-P1) for early detections of bone infections and evaluated the role of pathogen-related factors in the appli¬cations of 18F-FDG PET in the diagnostics of bone infections. For preclinical models of S. epidermidis and S. aureus bone/implant infections, the significance of the causative pathogen was studied with respect to 18F-FDG uptake. In a retrospective analysis of patients with confirmed bone infections, the significance of the presence or absence of positive bacterial cultures on 18F-FDG uptake was evalu¬ated. 18F-FDG and 68Ga-chloride resulted in a similar uptake in S. aureus osteomyelitic bones. However, 68Ga-chloride did not show uptake in healing bones, and therefore it may be a more-specific tracer in the early post-operative or post-traumatic phase. 68Ga-DOTAVAP-P1, a novel synthetic peptide bind¬ing to vascular adhesion protein 1 (VAP-1), was able to detect the phase of inflammation in healing bones, but the uptake of the tracer was elevated also in osteomyelitis. Low-grade peri-implant infec¬tions due to S. epidermidis were characterized by a low uptake of 18F-FDG, which reflects the virulence of the causative pathogen and the degree of leukocyte infiltration. In the clinical study, no relationship was found between the level of 18F-FDG uptake and the presence of positive or negative bacterial cul¬tures. Thus 18F-FDG PET may help to confirm metabolically active infection process in patients with culture-negative, histologically confirmed, low-grade osteomyelitis.

Save Consumers Time and Money: Thou Shall Not Forget Digital Native Big Data Consumers

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Office File Formats – What’s to be Done?

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Building Dynamic Data Centers for Fast Delivery of New Data and Data Updates

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Digital Preservation Tools, Practices, and Policies in Islandora

Presentation at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Dialysis, time and death: comparisons of two consecutive decades among patients treated at the same Brazilian dialysis center

The survival of hemodialysis patients is likely to be influenced not only by well-known risk factors like age and comorbidity, but also by changes in dialysis technology and practices accumulated along time. We compared the survival curves, dialysis routines and some risk factors of two groups of patients admitted to a Brazilian maintenance hemodialysis program during two consecutive decades: March 1977 to December 1986 (group 1, N = 162) and January 1987 to June 1997 (group 2, N = 237). The median treatment time was 22 months (range 1-198). Survival curves were constructed using the Kaplan-Meier method and compared using the log-rank method. The Cox proportional hazard regression model was used to investigate the more important variables associated with outcome. The most important changes in dialysis routine and in patient care during the total period of observation were the progressive increase in the dose of dialysis delivered, the prohibition of potassium-free dialysate, the use of bicarbonate as a buffer and the upgrading of the dialysis equipment. There were no significant differences between the survival curves of the two groups. Survival rates at 1, 5 and 10 years were 84, 53 and 29%, respectively, for group 1 and 77, 42 and 21% for group 2. Patients in group 1 were younger (45.5 ± 15.2 vs 55.2 ± 15.9 years, P<0.001) and had a lower prevalence of diabetes (11.1 vs 27.4%, P<0.001) and of cardiovascular disease (9.3 vs 20.7%, P<0.001). According to the Cox multivariate model, only age (hazard ratio (HR) 1.04, confidence interval (CI) 1.03-1.05, P<0.001) and diabetes (HR 2.55, CI 1.82-3.58, P<0.001) were independent predictors of mortality for the whole group. Patients of group 2 had a lower prevalence of sudden death (19.1 vs 9.7%, P<0.001). After adjusting for age, diabetes and other mortality risk factors, the risk of death was 17% lower in group 2, although this difference was not statistically significant. We conclude that the negative effects of advanced age and of higher frequency of comorbidity on the survival of group 2 patients were probably offset by improvements in patient care and in the quality and dose of dialysis delivered, so that the survival curves did not undergo significant changes along time.