Harsh winter climate causes various problems for both the public andprivate sector in Sweden, especially in the northern part and the railway industryis no exception. This master thesis project covers an investigation of the eects ofthe winter climate and a phenomena called atmospheric icing on the performance ofthe train in a region called the Botnia-Atlantica region. The investigation was donewith data over a short period January-February 2017 with simulated weather datafrom the Weather research and forecast model that was compared with the periodOctober - December 2016. The investigation only included high speed trains.The trains have been analysed based on two dierent performance measurements.The cumulative delay which is the increment in delay over a section and the currentdelay which is the current delay compared to the schedule. Cumulative delaysare investigated with survival analysis and the current delay is investigated with aMulti-state Markov model.The results show that the weather could have an eect on the trains performancewhere the survival analysis detected connection between the weather and cumulativedelays. The Markov model also showed a connection between the weather anddelayed trains including that the presence of atmospheric icing had a negative eecton remaining in a state of non-delay.

Doctor of Philosophy / With the fast development of artificial intelligence (AI) technology, the reliability of AI needs to be investigated for confidently using AI products in our daily lives. This dissertation includes three projects introducing the statistical models and model estimation methods that can be used in the reliability analysis of AI systems. The first project analyzes the recurrent events data from autonomous vehicles (AVs). A nonparametric model is proposed to study the reliability of AI systems in AVs, and a statistical framework is introduced to evaluate the adequacy of using traditional parametric models in the analysis. The proposed model and framework are then applied to analyze AV data from four manufacturers that participated in an AV driving testing program overseen by the California Department of Motor Vehicles. The second project develops a survival model to investigate the failure times of graphics processing units (GPUs) used in supercomputers. The model considers several covariates, the spatial correlation, and the correlation among multiple types of failures. In addition, unique spatial correlation functions and a special distance function are introduced to quantify the spatial correlation inside supercomputers. The model is applied to explore the GPU failure times in the Titan supercomputer. The third project proposes a new Markov chain Monte Carlo sampler that can be used in the estimation and inference of spatial survival models. The sampler can generate a reasonable amount of samples within a shorter computing time compared with existing popular samplers. Important factors that can influence the performance of the proposed sampler are explored, and the sampler is used to analyze the Titan GPU failures to illustrate its usefulness in solving real-world problems.

在過去文獻中多有討論遷徙行為者，但少有對家戶的居住時間進行分析，本研究首先檢視自有者與租賃者之居住時間，其次為瞭解家戶對實質住宅的經濟需求較大，還是對已建立的社會關係網絡需求較大，提出居住時間的時間相依性問題，即隨著居住時間的增加，家戶會因為對住宅需求改變而愈容易遷徙？還是在累積惰性下愈不易遷徙？最後則是探討家戶居住時間的影響因素，藉由不同的住宅與家戶特徵來瞭解家戶可能的居住時間，以作為住宅遷徙、流動性及其他住宅政策分析之參考。 本研究在實證方法上採取存活分析法，並利用華人家庭動態資料庫的橫斷面跨時資料進行研究。實證結果顯示，自有者的存活函數明顯大於租賃者，且自有家戶之居住時間為租賃家戶的3.11倍。另外，家戶遷徙機率具有正時間相依性，反應出家戶隨著居住時間的增加，遷徙機率會增加，然而自有家戶模型的遷徙機率卻呈現負時間相依性。居住時間影響因素實證結果，每人坪數較大，居住時間會較短，而年齡愈大與居住人數愈多者，居住時間會較長。 / Although there are lots of studies exploring migration, few of them focus on analysis household duration of residence. Duration of residence is an important source of information for a variety of private and pubic decisions and for the understanding of various social and economic phenomena. This study makes a residence duration comparison between owners and tenants. Due to cumulative stress and inertia, moving probability will be increased or decreased. We attempt to describe the duration of residence is positive or negative time dependent. This study also provides the evidence that house and household characteristics determine the duration of residence. This empirical study employs the survival analysis and uses the data from Panel Study of Family Dynamics. The result indicates that the survival function of owners is larger than of tenant. Duration of owner residence is 3.11 times longer than that of tenants. Generally, household moving probability is positive while the owner’s is negative. Research finding further reveals that ping per person is negatively related to residence duration. Age and household size are positively related to residence duration.

Approved for public release, distribution is unlimited / The goal of this thesis is to identify and quantitatively evaluate the factors, especially commissioning source, that affect the longevity of officers in the U.S. Navy. To reach this goal, a survival analysis is conducted on the survival patterns of officer cohorts who entered the service between the years 1983 and 1990. Using data created from Navy Officer Data Card information and annual promotion board results, three survival analysis procedures, LIFETEST, LIFEREG and PHREG were used to examine the factors that influence the survival of U.S. Naval Officers. The results of the survival analysis indicate that commissioning source has significant strong effect on survival rates with Naval Academy graduates have a better survival rate than other commissioning sources. Also, the analysis show that females and African-Americans have better survival rates than males and whites, respectively, and prior enlisted, older, graduates from non-selective colleges have higher survival rates than their counterparts. Additionally, Surface Warfare, Fleet Support and Supply Corps officers were found to have lower survival rates than officers in other communities. When survival functions for involuntary and voluntary separations were analyzed separately, the results were found different. Commissioning age, being African- American, single with children, commissioned from NROTC Contract Program, commissioned from OTHERSOURCE, being prior enlisted, having high GPAs and designated in AIR community had significant, negative effects on involuntary separations and significant positive effects on voluntary separations. / Captain, Turkish General Command of Gendarmerie

<p dir="rtl" style="text-align: left;">Karcinom bronha je važna i &scaron;iroko rasprostranjena bolest koja predstavlja veliki problem javnog zdravlja. Kod osoba mu&scaron;kog pola se nalazi na prvom mestu kako po učestalosti obolevanja tako i kao uzrok smrti među svim malignim tumorima, dok se kod osoba ženskog pola nalazi na trećem ili četvrtom mestu po obolevanju, a po umiranju uglavnom na drugom mestu. Kod najvećeg broja bolesnika bolest se otkriva u uznapredovalom ili metatstaskom stadijumu, a hemioterapija predstavlja jedan od vidova lečenja uznapredovale ili metastatske bolesti. Pored toga &scaron;to produžava preživljavanje i pobolj&scaron;ava kvalitet života obolelih praćena je istovremeno i brojnim neželjenim događajima. Iako hemioterapijski protokoli bazirani na preparatima platine ostvaruju najveću efektivnost kod bolesnika sa uznapredovalim nemikrocelularnim karcinomom bronha toksičnost koja je prati predstavlja najveći problem sa kojim se susrećemo. Ali, uprkos ovoj činjenici veliki broj studija je pokazao da je upravo odsustvo mijelotoksičnosti tokom hemioterapije udruženo sa lo&scaron;ijim ishodom lečenja kod obolelih od karcinoma bronha. Iz tog razloga je i predloženo da se hematolo&scaron;ka toksičnost koristi kao mera biolo&scaron;ke aktivnosti citotoksičnih lekova, njen prognostički značaj je evaluiran i proučavan u velikom broju studija. Ciljevi ove doktorske disertacije su bili da se utvrditi uticaj leukopenije, anemije i trombocitopenije kao nezavisnih prognostičkih faktora na preživljavanje bolesnika sa nemikrocelularnim karcinomom bronha; da se utvrditi učestalost hematolo&scaron;ke toksičnosti lečenih hemioterapijskim protokolima Cisplatin/Vepezid i Gemcitabin/Cisplatin i da se utvrdite razlike u preživljavanju bolesnika lečenih hemioterapijskim protokolima Cisplatin/Vepezid i Gemcitabin/Cisplatin. U uzorku je analizirano 200 bolesnika, 76% mu&scaron;kog i 24% ženskog pola, prosečne starosti 61.4 godine. Najzastupljenji su bili bolesnici u IV stadijumu 50.5%, a najče&scaron;ći patohistolo&scaron;ki tip karcinoma u uzorku je bio adenokarcinom sa 51.5%. Nije utvrđeno postojanje statistički značajne razlike u gradusima leukopenije, anemije i trombocitopenije između posmatranih grupa, (<em>X</em>²=2.908, <em>X</em>²=2.264, <em>X</em>²=3.403, p&gt;0.05). U obe grupe je univarijantnom analizom dokazanao da stadijum bolesti i terapijski odgovor imaju statistički značaj kao &bdquo;ne - hematolo&scaron;ki&ldquo; prognostički faktori (p&lt;0.01). U obe grupe su takođe univarijentnom analizom leukopenija, anemija i trombocitopenija identifikovane kao prognostički faktori kod obolelih od NSCLC, dok multivarijantnom analizom ni jedan od analiziranih faktora nije identifikovan kao prognostički. U obe grupe su bolesnici sa leukopenijom, anemijom i trombocitopenijom gradusa 3 i 4 imali statistički značajno duže preživljavanje u odnosu na bolesnike sa gradusom 0. Nije postojala razlika u preživljavanju bolesnika lečenih hemioterapijskim protokolima cisplatin/etopozid I gemcitabin/cisplatin (F=0.069; p&gt;0.05). Nije postojala razlika u preživljavanju bolesnika sa anemijom, leukopenijom i trombocitopenijom između grupa A i B za graduse 0, 3 i 4</p> / <p>Lung cancer is an important and widespread disease which represents a major public health problem. It is the most frequent disease among all malignant diseases at males, among women it is on the third or fourth place among malignant diseases. In most cases the disease is detected at an advanced or metastatic stage and chemotherapy is one of the therapy options of. Despite the fact that chemotherapy prolongs survival and improves quality of life of patients, at the same time chemotherapy causes a number of different adverse events. Although chemotherapy protocols based platinum achieve maximum effectiveness in patients with advanced non-small cell lung cancer, toxicity that accompanies represents a big problem. But despite this fact, a number of studies have shown that the absence of myelotoxicity during chemotherapy is associated with poorer treatment outcomes in patients with bronchial carcinoma. For this reason, it is proposed that hematological toxicity may be used as a measure of the biological activity of the cytotoxic drug, and its prognostic significance was studied and evaluated in a number of studies. The objectives of this dissertation were to determine the effect of leucopenia, anemia andthrombocytopenia as an independent prognostic factor in the survival of patients with Non Small Cell Lung lung cancer, to determine the incidence of hematological toxicity treated with chemotherapy protocols cisplatin/etoposid and gemcitabine/cisplatin and to determine differences in survival patients treated with chemotherapy protocols cisplatin/etoposid and gemcitabine/ cisplatin. We analyzed 200 patients, 76% male and 24% female, mean age 61.4 years. The most frequent were patients in stage IV 50.5%, and the most common histopathological type was adenocarcinoma with 51.5%. We did not find statistically significant differences in grade of leukopenia, anemia and thrombocytopenia between the groups (<em>X</em>²=2.908,<em> X</em>²=2.264, <em>X</em>²=3.403, p&gt;0.05). In both groups, the univariant analysis has shown that the stage of disease and response rate as a non-hematological prognostic factor had statistical significance (p &lt;0.01). In both groups of patients with NSCLC leucopenia, anemia and trombocitopenia has identified with univariant analysis as a prognostic factors, but multivariant analysis did not show that any of analyzed factors are prognostic. In both groups, patients with grade 3 and 4 of leucopenia, anemia and trombocitopenia had statistically longer survival than patients with grade 0. We did not find statistically significant difference in overall survival of patients treated with cisplatin/etoposid i gemcitabin/cisplatin regimes (F=0.069; p&gt;0.05). We did not find any statistically differences in overall survival between group A and B for leucopenia, anemia and trombocitopenia grade 0, 3 and 4.</p>

The topic of badgers in the UK is often a contentious one, dividing opinions and sparking political debate. On one hand, badgers represent an important part of the British ecosystem but on the other a wildlife reservoir of disease implicated in the transmission of bovine tuberculosis (TB) to livestock in the UK. This has prompted strong interest in their population dynamics and epidemiology. Using data from a long-term study of a naturally infected badger population in Woodchester Park, Gloucestershire, this thesis explores a range of capture-mark-recapture (CMR) models to further understand disease and demographic processes. The first section examines long term population dynamics, simultaneously estimating demographic rates alongside their drivers using integrated population models (IPMs). The findings provide new insight into badger demography, highlighting density-dependent mechanisms, vulnerabilities to changing climate and disease prevalence and subsequently how multi-factorial analyses are required to explain fluctuating badger populations. The following sections use multistate models to answer pertinent questions regarding individual disease dynamics, revealing rates of TB infection, progression and disease-induced mortality. A key finding was sex-differences in disease response, with males more susceptible to TB infection. After applying a survival trajectory analysis we suggest sex differences are due to male immune defence deficiencies. A comparative analysis demonstrated similarities between epidemiological processes at Woodchester Park to an unconnected population of badgers from a vaccine study, supporting its continued use as a model population. The final study in this thesis constructs an IPM to estimate disease and population dynamics and in doing so uncovers disease-state recruitment allocation rates, demographic and population estimates of badgers in varying health-states and predicts future dynamics. This model aims to encapsulate the more commonly held notion of populations as dynamic entities with numerous co-occurring processes, opening up avenues for future analyses within both the badger-TB system and possible extensions to other wildlife reservoir populations.

Em análise de sobrevivência estuda-se o tempo até a ocorrência de um determinado evento de interesse e na literatura, a abordagem mais usual é a paramétrica, em que os dados seguem uma distribuição de probabilidade. Diversas distribuições conhecidas são utilizadas para acomodar dados de tempos de falha, porém, grande parte destas distribuições não é capaz de acomodar funções de risco não monótonas. Kumaraswamy (1980) propôs uma nova distribuição de probabilidade e, baseada nela, mais recentemente Cordeiro e de Castro (2011) propuseram uma nova família de distribuições generalizadas, a Kumaraswamy generalizada (Kum-G). Esta distribuição além de ser flexível, contém distribuições com funções de risco unimodal e em forma de banheira. O objetivo deste trabalho é apresentar a família de distribuições Kum-Ge seus casos particulares para analisar dados de tempo de vida dos indivíduos em risco, considerando que uma parcela da população nunca apresentará o evento de interesse, além de considerarmos que covariáveis influenciem na função de sobrevivência e na proporção de curados da população. Algumas propriedades destes modelos serão abordadas, bem como métodos adequa- dos de estimação, tanto na abordagem clássica quanto na bayesiana. Por fim, são apresentadas aplicações de tais modelos a conjuntos de dados existentes na literatura. / In survival analysis is studied the time until the occurrence of a particular event of interest and in the literature, the most common approach is parametric, that the data follow a probability distribution. Various known distributions are used to accommodate failure times data, however, most of these distributions is not able to accommodate non monotonous hazard functions. Kumaraswamy (1980) proposed a new probability distribution and, based on it, most recently Cordeiro e de Castro (2011) proposed a new family of generalized distributions, Kumaraswamy generalized (Kum-G). This distribution besides being flexible, has distributions with unimodal and tub form of hazard functions. The objective of this paper is to present the family of Kum-G distributions and their particular cases to analyze lifetime data of individuals at risk, considering that part of the population never present the event of interest, and considering that covariates influencing in the survival function and the cured proportion of the population. Some properties of these models will be discussed as well as appropriate estimation methods, in the classical and Bayesian approaches. Finally, applications of such models are presented to data sets existingin the literature.

Propensity score methods are increasingly being used to reduce the effect of measured confounders in observational research. In medicine, censored time-to-event data is common. Using Monte Carlo simulations, this thesis evaluates the performance of nearest neighbour matching (NNM) and inverse probability of treatment weighting (IPTW) in combination with Cox proportional hazards models for estimating marginal hazard ratios. Focus is on the performance for different sample sizes and censoring rates, aspects which have not been fully investigated in this context before. The results show that, in the absence of censoring, both methods can reduce bias substantially. IPTW consistently had better performance in terms of bias and MSE compared to NNM. For the smallest examined sample size with 60 subjects, the use of IPTW led to estimates with bias below 15 %. Since the data were generated using a conditional parametrisation, the estimation of univariate models violates the proportional hazards assumption. As a result, censoring the data led to an increase in bias.

INTRODUÇÃO: As hérnias de parede abdominal têm elevada incidência em pacientes cirróticos. Ascite e desnutrição contribuem para que essas hérnias adquiram grandes proporções, causando sintomas álgicos, decréscimo da qualidade de vida e maior risco de complicações locais que exijam tratamento cirúrgico de urgência. Contudo, o tratamento conservador é o mais empregado pela elevada morbimortalidade associada a procedimentos cirúrgicos nesses pacientes. OBJETIVO: O objetivo deste estudo é analisar os resultados do tratamento cirúrgico de hérnias de parede abdominal em pacientes cirróticos. MÉTODOS: Estudo prospectivo, baseado no seguimento de pacientes cirróticos com hérnia de parede abdominal no Ambulatório de Transplante Hepático do Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo, durante o período de 01/01/2009 à 01/11/2014. Foram analisadas as características demográficas, da hepatopatia, do tipo de hérnia, ocorrência de complicações e mortalidade dos pacientes que foram submetidos à cirurgia eletiva ou mantidos em acompanhamento clínico. Desse último grupo, os pacientes que apresentaram complicações com indicação de cirurgia, foram submetidos à cirurgia de urgência. Os pacientes submetidos à cirurgia eletiva foram selecionados de forma randômica. RESULTADOS: Foram avaliados 246 pacientes nesse período do estudo. A cirurgia eletiva foi realizada em 57 pacientes. Desses, 186 permaneceram em acompanhamento clínico. A incidência de complicações exigindo tratamento cirúrgico de urgência foi de 22,7% (43 pacientes). A sobrevida a longo prazo foi maior entre os pacientes submetidos à cirurgia eletiva em relação aos pacientes em acompanhamento clínico (p=0,012). A cirurgia de urgência apresentou maior incidência de complicações pós-operatórias e maior mortalidade em comparação à cirurgia eletiva (p=0,005). CONCLUSÕES: O paciente cirrótico, com hérnia de parede abdominal, apresenta elevada mortalidade em sua evolução, independente da realização da correção cirúrgica da hérnia. A cirurgia eletiva proporcionou maior sobrevida em comparação aos pacientes mantidos em acompanhamento clínico. A cirurgia de urgência foi um fator de risco para maior morbidade e mortalidade / INTRODUCTION: Cirrhotic patients have higher incidence of abdominal wall hernias. Ascistes and sarcopenia are risk factors to development of huge hernias, leading to pain, poor quality of life and need of urgent surgery due local complications. However, hernia surgery is usually delayed among cirrhotic patients due higher morbidity and mortality. OBJECTIVE: This study aimed to analyze the surgical treatment of abdominal wall hernias in cirrhotic patients. METHODS: A prospective, analytical study, based on follow-up of cirrhotic patients with abdominal wall hernia that were followed in Liver Transplant Clinic at the Hospital das Clinicas, Faculty of Medicine, University of São Paulo, during the period from January 2009 to November 2014. We analyzed demographics, characteristics of liver disease, type of hernia, complications and mortality of patients who underwent elective hernia surgery or were kept in exclusive clinical follow up. The exclusive clinical group underwent urgent hernia surgery when indicated. Elective surgery was performed in unselected patients. RESULTS: We enrolled 246 patients during the study period. Elective surgery was performed in 57 patients. 186 patients had clinical follow up, incidence of urgent surgery was 22,7% (43 patients). Elective surgery provided better long term survival then clinical follow up (p=0.012). Urgent surgery had higher morbidity and it was an independent risk factor for mortality (p=0.005). CONCLUSIONS: cirrhotic patients with abdominal wall hernia have higher mortality, regardless of performing the surgical correction. Hernia elective repair provided better survival than patients in conservative treatment. Urgent surgery imposes higher morbidity and mortality than elective surgery

Censura intervalar resulta quando os tempos de sobrevivência não são exatamente conhecidos, sabe-se apenas que eles ocorreram dentro de um intervalo. Dados de sobrevivência agrupados são casos particulares de censura intervalar quando os indivíduos são avaliados nos mesmos intervalos de tempo, ocasionando um grande número de empates. Um procedimento comum para a análise desse tipo de dados é ignorar a natureza de censura intervalar dos dados, ou seja, tratar a variável aleatória tempo como contínua e assumir que o evento ocorreu no início, no ponto médio ou no final do intervalo e, então, usar um método padrão de análise de sobrevivência. Neste estudo, simulações de Monte Carlo, com o modelo de Weibull, foram realizadas para comparar esses três procedimentos e um método novo proposto que é uma combinação desses três métodos e é orientado pela observação do histograma do tempo versus a freqüência de cada intervalo para a decisão de qual valor a ser usado. Considera-se também a análise dos dados como censura intervalar. Os resultados mostram que analisar os dados exatamente como censura intervalar é a forma correta. Entretanto, quando a taxa de falha aumenta o ponto médio poderia ser usado. A natureza discreta dos tempos de falha deve ser reconhecida quando existe um grande número de empates. Métodos de regressão para tratar dados agrupados são apresentados por Lawless (2003) e Collett (2003), cuja estrutura é especificada em termos da probabilidade de um indivíduo falhar em um intervalo, condicionada à sua sobrevivência ao intervalo anterior. Os modelos considerados na literatura são o de riscos proporcionais de Cox ou o logístico. O modelo de Weibull é proposto, neste trabalho, como uma alternativa ao modelo de Cox para ajustar dados de sobrevivência com censura intervalar no contexto de modelos discretos. Através de simulações foram construídas as estatísticas da razão de verossimilhança e do teste escore para a discriminação entre esses dois modelos. Para ilustrar as simulações duas aplicações em dados agronômicos foram utilizadas. / Interval-censored results when survival times are not exactly known, knowing only that they occur in an interval. Grouped survival data are particular cases of intervalcensored when individuals are evaluated in the same time-intervals, causing a great number of ties. A common procedure for the analysis this type of data is to ignore the nature of interval-censored data, or rather, treat the random variable time as continuous, and assume that the event occurred in the beginning, midpoint or interval end, and then use a standard method of survival analysis. In this study, Monte Carlo simulations according to Weibull model, were performed in order to compare these three procedures and a new method proposed which is a combination of the three, and is directed by the observation of time histogram versus each interval frequency in order to decide which value be used. Interval-censored data is also considered. The results show that to analyse the data exactly as interval-censored is the correct form. However, when the failure rate increase the midpoint could be used. The discrete nature of failure time must be recognized when there are a great number of ties. Regression methods to treat grouped data are presented by Lawless (2003) and Collett (2003), whose structure is specified in terms of the probability of an individual failing in an interval, conditioned to his survival to previous interval. The models considered in literature are either those of Cox proportional hazards or the logistic one. Weibull model is proposed in this study as an alternative to Cox model in order to adjust survival data with interval-censored in the context of discrete models. Through simulations were built the statistics of ratio likelihood and score test to distinguish between these two models. To illustrate the simulations two applications in agronomy data were used.