On probability-based inference under data missing by design

Whether a statistician wants to complement a probability model for observed data with a prior distribution and carry out fully probabilistic inference, or base the inference only on the likelihood function, may be a fundamental question in theory, but in practice it may well be of less importance if the likelihood contains much more information than the prior. Maximum likelihood inference can be justified as a Gaussian approximation at the posterior mode, using flat priors. However, in situations where parametric assumptions in standard statistical models would be too rigid, more flexible model formulation, combined with fully probabilistic inference, can be achieved using hierarchical Bayesian parametrization. This work includes five articles, all of which apply probability modeling under various problems involving incomplete observation. Three of the papers apply maximum likelihood estimation and two of them hierarchical Bayesian modeling. Because maximum likelihood may be presented as a special case of Bayesian inference, but not the other way round, in the introductory part of this work we present a framework for probability-based inference using only Bayesian concepts. We also re-derive some results presented in the original articles using the toolbox equipped herein, to show that they are also justifiable under this more general framework. Here the assumption of exchangeability and de Finetti's representation theorem are applied repeatedly for justifying the use of standard parametric probability models with conditionally independent likelihood contributions. It is argued that this same reasoning can be applied also under sampling from a finite population. The main emphasis here is in probability-based inference under incomplete observation due to study design. This is illustrated using a generic two-phase cohort sampling design as an example. The alternative approaches presented for analysis of such a design are full likelihood, which utilizes all observed information, and conditional likelihood, which is restricted to a completely observed set, conditioning on the rule that generated that set. Conditional likelihood inference is also applied for a joint analysis of prevalence and incidence data, a situation subject to both left censoring and left truncation. Other topics covered are model uncertainty and causal inference using posterior predictive distributions. We formulate a non-parametric monotonic regression model for one or more covariates and a Bayesian estimation procedure, and apply the model in the context of optimal sequential treatment regimes, demonstrating that inference based on posterior predictive distributions is feasible also in this case.

Problems and Algorithms for Sequence Segmentations

The analysis of sequential data is required in many diverse areas such as telecommunications, stock market analysis, and bioinformatics. A basic problem related to the analysis of sequential data is the sequence segmentation problem. A sequence segmentation is a partition of the sequence into a number of non-overlapping segments that cover all data points, such that each segment is as homogeneous as possible. This problem can be solved optimally using a standard dynamic programming algorithm. In the first part of the thesis, we present a new approximation algorithm for the sequence segmentation problem. This algorithm has smaller running time than the optimal dynamic programming algorithm, while it has bounded approximation ratio. The basic idea is to divide the input sequence into subsequences, solve the problem optimally in each subsequence, and then appropriately combine the solutions to the subproblems into one final solution. In the second part of the thesis, we study alternative segmentation models that are devised to better fit the data. More specifically, we focus on clustered segmentations and segmentations with rearrangements. While in the standard segmentation of a multidimensional sequence all dimensions share the same segment boundaries, in a clustered segmentation the multidimensional sequence is segmented in such a way that dimensions are allowed to form clusters. Each cluster of dimensions is then segmented separately. We formally define the problem of clustered segmentations and we experimentally show that segmenting sequences using this segmentation model, leads to solutions with smaller error for the same model cost. Segmentation with rearrangements is a novel variation to the segmentation problem: in addition to partitioning the sequence we also seek to apply a limited amount of reordering, so that the overall representation error is minimized. We formulate the problem of segmentation with rearrangements and we show that it is an NP-hard problem to solve or even to approximate. We devise effective algorithms for the proposed problem, combining ideas from dynamic programming and outlier detection algorithms in sequences. In the final part of the thesis, we discuss the problem of aggregating results of segmentation algorithms on the same set of data points. In this case, we are interested in producing a partitioning of the data that agrees as much as possible with the input partitions. We show that this problem can be solved optimally in polynomial time using dynamic programming. Furthermore, we show that not all data points are candidates for segment boundaries in the optimal solution.

Optimisation Problems in Wireless Sensor Networks : Local Algorithms and Local Graphs

This thesis studies optimisation problems related to modern large-scale distributed systems, such as wireless sensor networks and wireless ad-hoc networks. The concrete tasks that we use as motivating examples are the following: (i) maximising the lifetime of a battery-powered wireless sensor network, (ii) maximising the capacity of a wireless communication network, and (iii) minimising the number of sensors in a surveillance application. A sensor node consumes energy both when it is transmitting or forwarding data, and when it is performing measurements. Hence task (i), lifetime maximisation, can be approached from two different perspectives. First, we can seek for optimal data flows that make the most out of the energy resources available in the network; such optimisation problems are examples of so-called max-min linear programs. Second, we can conserve energy by putting redundant sensors into sleep mode; we arrive at the sleep scheduling problem, in which the objective is to find an optimal schedule that determines when each sensor node is asleep and when it is awake. In a wireless network simultaneous radio transmissions may interfere with each other. Task (ii), capacity maximisation, therefore gives rise to another scheduling problem, the activity scheduling problem, in which the objective is to find a minimum-length conflict-free schedule that satisfies the data transmission requirements of all wireless communication links. Task (iii), minimising the number of sensors, is related to the classical graph problem of finding a minimum dominating set. However, if we are not only interested in detecting an intruder but also locating the intruder, it is not sufficient to solve the dominating set problem; formulations such as minimum-size identifying codes and locating dominating codes are more appropriate. This thesis presents approximation algorithms for each of these optimisation problems, i.e., for max-min linear programs, sleep scheduling, activity scheduling, identifying codes, and locating dominating codes. Two complementary approaches are taken. The main focus is on local algorithms, which are constant-time distributed algorithms. The contributions include local approximation algorithms for max-min linear programs, sleep scheduling, and activity scheduling. In the case of max-min linear programs, tight upper and lower bounds are proved for the best possible approximation ratio that can be achieved by any local algorithm. The second approach is the study of centralised polynomial-time algorithms in local graphs these are geometric graphs whose structure exhibits spatial locality. Among other contributions, it is shown that while identifying codes and locating dominating codes are hard to approximate in general graphs, they admit a polynomial-time approximation scheme in local graphs.

Confronting Confucian Understandings of the Christian Doctrine of Salvation : A systematic theological analysis of the basic problems in the Confucian-Christian dialogue

The aim of the present study is to analyze Confucian understandings of the Christian doctrine of salvation in order to find the basic problems in the Confucian-Christian dialogue. I will approach the task via a systematic theological analysis of four issues in order to limit the thesis to an appropriate size. They are analyzed in three chapters as follows: 1. The Confucian concept concerning the existence of God. Here I discuss mainly the issue of assimilation of the Christian concept of God to the concepts of Sovereign on High (Shangdi) and Heaven (Tian) in Confucianism. 2. The Confucian understanding of the object of salvation and its status in Christianity. 3. The Confucian understanding of the means of salvation in Christianity. Before beginning this analysis it is necessary to clarify the vast variety of controversies, arguments, ideas, opinions and comments expressed in the name of Confucianism; thus, clear distinctions among different schools of Confucianism are given in chapter 2. In the last chapter I will discuss the results of my research in this study by pointing out the basic problems that will appear in the analysis. The results of the present study provide conclusions in three related areas: the tacit differences in the ways of thinking between Confucians and Christians, the basic problems of the Confucian-Christian dialogue, and the affirmative elements in the dialogue. In addition to a summary, a bibliography and an index, there are also eight appendices, where I have introduced important background information for readers to understand the present study.

Postoperative soft tissue defects in the ankle area : The etiology and methods of reconstruction

The risk is obvious for soft tissue complications after operative treatment of the Achilles tendon, calcaneal bone or after ankle arthroplasty. Such complications after malleolar fractures are, however, seldom seen. The reason behind these complications is that the soft tissue in this region is tight and does not allow much tension to the wound area after surgery. Furthermore the area of operation may be damaged by swelling after the injury, or can be affected by peripheral vascular disease. While complications in this area are unavoidable, they can be diminished. This study attempts to highlight the possible predisposing factors leading to complications in these operations and on the other hand, to determine the solutions to solve soft tissue problems in this region. The study consists of five papers. The first article is a reprint on the soft tissue reconstruction of 25 patients after their complicated Achilles tendon surgeries were analysed. The second study reviews a series of 126 patients after having undergone an operative treatment of calcaneal bone fractures and analyses the complications and possible reasons behind them. The third part analyses a series of corrections of 35 soft tissue complications after calcaneal fracture operations. The fourth part reviews a series of 7 patients who had undergone complicated ankle arthroplasties. The last article presents a series of post operative lateral defects of the ankle treated with a less frequently used distally based peroneus brevis muscle flap and analyses the results. What can be conducted from these studies is that in general, the results after the correction of even severe soft tissue complications in the ankle region are good. For the small defects around the Achilles tendon, the local flaps are useful, but the larger defects are best treated with a free flap. We found that a long delay from trauma to surgery and a long operating time were predisposing factors that lead to soft tissue complications after operatively treated calcaneal bone fractures. The more severe the injury, the greater the risk for wound complication. Surprisingly, the long-term results after infected calcaneal osteosyntheses were acceptable and the calcaneal bone seems to tolerate chronic infections very well if the soft tissue is reconstructed successfully. Behind the complicated ankle arthroplasties, unexpectedly high number of cases experiencing arteriosclerosis of the lower extremity was found. These complications lead to ankle fusion but can be solved with a free flap if the vascularity is intact or can be reconstructed. For this reason a vascular examination of the lower extremity arteries of the patients going to ankle arthroplasty is strongly recommended. Moreover postoperative lateral malleolar wound infections which typically create lateral ankle defects can successfully be treated with a peroneus brevis muscle flap covered with a free skin graft.

Prevalence and determinants of respiratory symptoms, asthma, chronic bronchitis and allergic sensitization in Helsinki : A comparison between Finland, Sweden and Estonia. The FinEsS studies - Helsinki I

Objectives: To assess the prevalence and risk factor profiles of respiratory symptoms, asthma and chronic bronchitis in Helsinki, and to compare these results with those for Sweden and Estonia. Other important aims were to evaluate the prevalence and determinants of type 1 sensitization in Helsinki. Materials and methods: This presentation is a part of a large epidemiological study in Finland, Estonia and Sweden (FinEsS). The first part of the study consisted of a postal questionnaire in 1995-1996 distributed to subjects in eight study centres. The study population in each centre was a population-based random sample designed to be representative of the general population. The original study sample in Helsinki consisted of 8000 subjects aged 20-69 years, 6062 (76%) of whom participated. Comparisons between countries were based on a narrower age group, 20-64 years, since 64 years was the upper age limit used in the original study in Estonia. Thus, altogether 58 661 subjects aged 20-64 years were invited to participate in Finland, Sweden and Estonia, and 44 483 (76%) did so. The second part of the study was a clinical study with a structured interview, lung function measurements and skin-prick tests with 15 common allergens. This thesis reports only the results of the prick tests in Helsinki. Of the 1200 subjects invited to participate in Helsinki, 643 (54%) consented. Skin-prick tests were performed on subjects ≤ 60 years of age; thus, a total of 498 tests were done. Results: In Helsinki, the prevalence of physician-diagnosed asthma was 6.6% and of physician-diagnosed chronic bronchitis 3.7% among subjects aged 20-69 years. Comparison of the results between Finland, Sweden and Estonia in subjects 20-64 years of age revealed the highest prevalence of physician-diagnosed asthma in Sweden, 7.8%, while the prevalence in Finland was 5.9% and in Estonia 2.0% (p<0.001). The prevalence of physician-diagnosed asthma among those aged 20-29 years was 7.9% in Stockholm, 6.3% in Helsinki and 2.8% in Tallinn. Asthma-related symptoms were most common in Estonia, and among those with typical asthma symptoms the diagnosis of asthma was least likely in Estonia. Physician-diagnosed chronic bronchitis was reported to be 10.7% in Estonia, 3.1% in Sweden and 2.9% in Finland among subjects aged 20-64 years (p<0.001). Among those aged 20-29 years, 7.6% in Tallinn reported physician-diagnosed chronic bronchitis, while the prevalence estimates were 1.4% in Stockholm and 1.3% in Helsinki. The prevalence of smoking was similar for women in all three countries, around 30%, but large differences in smoking habits were present among men; 60% of Estonian, 39% of Finnish and 28% of Swedish men smoked. Skin-prick tests in Helsinki revealed a high prevalence of sensitization, 46.9%. For subjects aged 26-39 years, the prevalence was highest, 56.8%, and 23.7% were sensitized to at least four allergens. The most common sensitizing allergen was the dog. Sensitization to multiple allergens was associated with a high prevalence of asthma and allergic rhinitis. Conclusions: Compared with earlier Finnish studies, a higher prevalence of asthma and a lower prevalence of chronic bronchitis were found in Helsinki. The prevalence of physician-diagnosed chronic bronchitis was low in Helsinki, with only one-fifth of subjects fulfilling the symptom criteria for chronic bronchitis reporting having a diagnosis of chronic bronchitis. The prevalences of asthma and chronic bronchitis were similar in Finland and Sweden, but in Estonia physician-diagnosed asthma was less common and physician-diagnosed chronic bronchitis more common, particularly among young subjects. Further analyses revealed that the diagnosis of asthma was favoured in Finland and Sweden, while the diagnosis of chronic bronchitis was more likely in Estonia for subjects with the same symptoms. Allergic sensitization was common in Helsinki. Our findings of multiple sensitization also speak in favour of evaluating the degree of sensitization when assessing allergies.

Clinical aspects of exhaled nitric oxide in adults : Associations with atopy, bronchial hyperresponsiveness, smoking and chronic obstruction

Measurement of fractional exhaled nitric oxide (FENO) has proven useful in assessment of patients with respiratory symptoms, especially in predicting steroid response. The objective of these studies was to clarify issues relevant for the clinical use of FENO. The influence of allergic sensitization per se on FENO in healthy asymptomatic subjects was studied, the association between airway inflammation and bronchial hyperresponsiveness (BHR) in steroid-naive subjects with symptoms suggesting asthma was examined, as well as the possible difference in this association between atopic and nonatopic subjects. Influence of smoking on FENO was compared between atopic and nonatopic steroid-naive asthmatics and healthy subjects. The short-term repeatability of FENO in COPD patients was examined in order to assess whether the degree of chronic obstruction influences the repeatability. For these purposes, we studied a random sample of 248 citizens of Helsinki, 227 army conscripts with current symptoms suggesting asthma, 19 COPD patients, and 39 healthy subjects. FENO measurement, spirometry and bronchodilatation test, structured interview. skin prick tests, and histamine and exercise challenges were performed. Among healthy subjects with no signs of airway diseases, median FENO was similar in skin prick test-positive and –negative subjects, and the upper normal limit of FENO was 30 ppb. In atopic and nonatopic subjects with symptoms suggesting asthma, FENO associated with severity of exercise- or histamine-induced BHR only in atopic patients. FENO in smokers with steroid-naive asthma was significantly higher than in healthy smokers and nonsmokers. Among atopic asthmatics, FENO was significantly lower in smokers than in nonsmokers, whereas no difference appeared among nonatopic asthmatics. The 24-h repeatability of FENO was equally good in COPD patients as in healthy subjects. These findings indicate that allergic sensitization per se does not influence FENO, supporting the view that elevated FENO indicates NO-producing airway inflammation, and that same reference range can be applied to both skin prick test-positive and -negative subjects. The significant correlation between FENO and degree of BHR only in atopic steroid-naive subjects with current asthmatic symptoms supports the view that pathogenesis of BHR in atopic asthma is strongly involved in NO-producing airway inflammation, whereas in development of BHR in nonatopic asthma other mechanisms may dominate. Attenuation of FENO only in atopic but not in nonatopic smokers with steroid-naive asthma may result from differences in mechanisms of FENO formation as well as in sensitivity of these mechanisms to smoking in atopic and nonatopic asthma. The results suggest, however, that in young adult smokers, FENO measurement may prove useful in assessment of airway inflammation. The short-term repeatability of FENO in COPD patients with moderate to very severe disease and in healthy subjects was equally good.

Systemic Inflammation in Preterm Infants

Even though mortality among preterm infants has decreased, their risk for chronic complications such as bronchopulmonary dysplasia (BPD) and neurological disability remains significant. One common risk factor for these is exposure to inflammation. The fetus may be exposed prenatally during maternal chorioamnionitis. Pre-eclampsia is also associated with low-grade maternal inflammation. Postnatally, local and systemic inflammation is present during respiratory distress syndrome (RDS). Furthermore, septic infections in the preterm infant are an important source of inflammatory stimuli and can lead to death in only a few hours. The diagnosis of septic infection is difficult, since reliable diagnostic markers are unavailable. This thesis evaluates peri- and postnatal systemic inflammation in preterm infants in septic infections, in RDS treated with mechanical ventilation and surfactant treatment, and in preterm infants prenatally exposed to chorioamnionitis and pre-eclampsia. Surface expressions of the activation markers CD11b, CD54, and CD62L, determined by flow cytometry on circulating phagocytes and T lymphocytes, serve as indicators of systemic inflammation. The main findings: I) In preterm infants with developing late-onset sepsis and fulminant necrotizing enterocolitis, a significant increase in CD11b expression on circulating phagocytes is already present on the day of onset of clinical symptoms. II) In preterm infants with RDS, circulating phagocytes become activated within hours after start of mechanical ventilation. In preterm infants treated for RDS with nasal continuous positive airway pressure, no such activation occurs. III) In preterm infants, RDS is associated during the first days of life with fewer circulating helper and cytotoxic T lymphocytes, of which the greater proportions are activated. Even greater proportions of circulating T cells are activated in infants subsequently developing BPD. IV) In preterm infants born after maternal pre-eclampsia, RDS-associated phagocyte CD11b up-regulation is greater than in preterm infants not exposed to pre-eclampsia during the first week of life. These findings suggest that I) an increase in CD11b expression on circulating phagocytes can identify preterm infants with late-onset sepsis as early as at sampling for blood culture and may thus aid in the diagnosis. II) In preterm infants with RDS, initiation of mechanical ventilation, but not the use of nasal continuous positive airway pressure, promotes a systemic inflammatory reaction; exogenous surfactant does not seem to promote inflammation. III) In addition to activation of circulating cells of the innate immunity in preterm infants with RDS, the circulating cells of the adaptive immunity are activated. The activation of adaptive immunity may link acute inflammation and development of chronic inflammation-associated problems such as BPD. IV) Maternal pre-eclampsia may prime neonatal immunity to react more strongly to postnatal stimuli. In conclusion, the preterm infant is exposed to numerous potentially injurious events such as intrauterine inflammation, respiratory distress syndrome (RDS), and systemic infections, all evoking systemic inflammation. Due to ongoing development of the lung and the brain, this may, in addition to acute injury, lead to aberrant lung and brain development and to clinical syndromes of BPD and neurologic sequelae.

Assessment of neuroleptic-induced movement disorders in a naturalistic schizophrenia population

The prevalence and assessment of neuroleptic-induced movement disorders (NIMDs) in a naturalistic schizophrenia population that uses conventional neuroleptics were studied. We recruited 99 chronic schizophrenic institutionalized adult patients from a state nursing home in central Estonia. The total prevalence of NIMDs according to the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) was 61.6%, and 22.2% had more than one NIMD. We explored the reliability and validity of different instruments for measuring these disorders. First, we compared DSM-IV with the established observer rating scales of Barnes Akathisia Rating Scale (BARS), Simpson-Angus Scale (SAS) (for neuroleptic-induced parkinsonism, NIP) and Abnormal Involuntary Movement Scale (AIMS) (for tardive dyskinesia), all three of which have been used for diagnosing NIMD. We found a good overlap of cases for neuroleptic-induced akathisia (NIA) and tardive dyskinesia (TD) but somewhat poorer overlap for NIP, for which we suggest raising the commonly used threshold value of 0.3 to 0.65. Second, we compared the established observer rating scales with an objective motor measurement, namely controlled rest lower limb activity measured by actometry. Actometry supported the validity of BARS and SAS, but it could not be used alone in this naturalistic population with several co-existing NIMDs. It could not differentiate the disorders from each other. Quantitative actometry may be useful in measuring changes in NIA and NIP severity, in situations where the diagnosis has been made using another method. Third, after the relative failure of quantitative actometry to show diagnostic power in a naturalistic population, we explored descriptive ways of analysing actometric data, and demonstrated diagnostic power pooled NIA and pseudoakathisia (PsA) in our population. A subjective question concerning movement problems was able to discriminate NIA patients from all other subjects. Answers to this question were not selective for other NIMDs. Chronic schizophrenia populations are common worldwide, NIMD affected two-thirds of our study population. Prevention, diagnosis and treatment of NIMDs warrant more attention, especially in countries where typical antipsychotics are frequently used. Our study supported the validity and reliability of DSM-IV diagnostic criteria for NIMD in comparison with established rating scales and actometry. SAS can be used with minor modifications for screening purposes. Controlled rest lower limb actometry was not diagnostically specific in our naturalistic population with several co-morbid NIMDs, but it may be sensitive in measuring changes in NIMDs.

Cytomegalovirus in the Process of Chronic Allograft Nephropathy

Cytomegalovirus (CMV) is a major cause of morbidity, costs and even mortality in organ transplant recipients. CMV may also enhance the development of chronic allograft nephropathy (CAN), which is the most important cause of graft loss after kidney transplantation. The evidence for the role of CMV in chronic allograft nephropathy is somewhat limited, and controversial results have also been reported. The aim of this study was to investigate the role of CMV in the pathogenesis of CAN. Material for the purpose of this study was available from altogether 70 kidney transplant recipients who received a kidney transplant between the years 1992-2000. CMV infection was diagnosed with pp65 antigenemia test or by viral culture from blood, urine, or both. CMV proteins were demonstrated in the kidney allograft biopsies by immunohistochemisrty and CMV-DNA by in situ hybridization. Cytokines, adhesion molecules, and growth factors were demonstrated from allograft biopsies by immunohistochemistry, and from urinary samples by ELISA-methods. CMV proteins were detectable in the 6-month protocol biopsies from 18/41 recipients with evidence of CMV infection. In the histopathological analysis of the 6-month protocol biopsies, presence of CMV in the allograft together with a previous history of acute rejection episodes was associated with increased arteriosclerotic changes in small arterioles. In urinary samples collected during CMV infection, excretion of TGF-β was significantly increased. In recipients with increased urinary excretion of TGF-β, increased interstitial fibrosis was recorded in the 6- month protocol biopsies. In biopsies taken after an active CMV infection, CMV persisted in the kidney allograft in 17/48 recipients, as CMV DNA or antigens were detected in the biopsies more than 2 months after the last positive finding in blood or urine. This persistence was associated with increased expression of TGF-β, PDGF, and ICAM-1 and with increased vascular changes in the allografts. Graft survival and graft function one and two years after transplantation were reduced in recipients with persistent intragraft CMV. Persistent intragraft CMV infection was also a risk factor for reduced graft survival in Cox regression analysis, and an independent risk factor for poor graft function one and two years after transplantation in logistic regression analysis. In conclusion, these results show that persistent intragraft CMV infection is detrimental to kidney allografts, causing increased expression of growth factors and increased vascular changes, leading to reduced graft function and survival. Effective prevention, diagnosis and treatment of CMV infections may a major factor in improving the long term survival of kidney allograft.