This thesis proposes the use of Machine Learning (ML) to two very distinct, yet compelling, applications – predicting cardiac arrhythmia episodes and predicting base station association in 5G networks comprising of virtual cells. In the first scenario, Support Vector Machines (SVMs) are used to classify features extracted from electrocardiogram (EKG) signals. The second problem requires a different formulation departing from traditional ML classification where the objective is to partition feature space into constituent class regions. Instead, the intention here is to identify temporal patterns in unequal-length sequences. Using Recurrent Neural Networks (RNNs), it is demonstrated that accurate predictions can be made as to the base station most likely to provide connectivity for a mobile device as it moves. Atrial Fibrillation (AF) is a common cardiac arrhythmia affecting several million people in the United States. It is a condition in which the upper chambers of the heart are unable to contract effectively leading to inhibited blood flow to the ventricles. The stagnation of blood is one of the major risk factors for stroke. The Computers in Cardiology Challenge 2001 was organized to further research into the prediction of episodes of AF. This research revisits the problem with some modifications. Patient-specific classifiers are developed for AF prediction using a different dataset and employing shorter EKG signal epochs. SVM classification yielded an average accuracy of just above 95% in identifying EKG epochs appearing just prior to fibrillatory rhythms. 5G cellular networks were envisaged to provide enhanced data rates for mobile broadband, support low-latency communication, and enable the Internet of Things (IoT). Handovers contribute to latency as mobile devices are switched between base stations due to movements. Given that customers may not be willing to continuously share their exact locations due to privacy concerns and the establishment of a mobile network architecture with dynamically created virtual cells, this research presents a solution for proactive mobility management using RNNs. A RNN is trained to identify patterns in variable-length sequences of Received Signal Strength (RSS) values, where a mobile device is permitted to connect to more than a single base station at a time. A classification accuracy of over 98% was achieved in a simulation model that was set up emulating an urban environment.

L'ensemble de ce travail est dédié à l'étude de certaines propriétés concernant les processus de sauts d-dimensionnels X = (Xt) dont le générateur est donné par Lψ(x) = 1/2 ∑ aᵤᵥ(x)∂²ψ(x)/∂xᵤ∂xᵥ + g(x)∇ψ(x) + ∫ (ψ(x + c(z, x)) − ψ(x))γ(z, x)µ(dz) où µ est de masse totale infinie. Si γ ne dépendait pas de x, nous nous trouverions dans une situation classique où le processus X pourrait être représenté comme une solution d'une équation stochastique comportant une mesure ponctuelle de Poisson de mesure d'intensité γ(z)µ(dz) ; lorsque γ dépend de x, on peut s'en représenter l'heuristique en imaginant le processus comme la trajectoire d'une particule, la loi des sauts pouvant alors dépendre de la position de la particule. Dans la première partie, nous donnons des conditions pour obtenir l'existence et l'unicité de tels processus. Ensuite, nous considérons ce type de processus comme une généralisation des PDMP ; nous montrons qu'ils peuvent être vus comme une limite d'une suite (Xᵣ(t)) de PDMP standards pour lesquels l'intensité des sauts tend vers l'infini quand r tend vers l'infini, suivant deux régimes : un lent et un rapide qui, en supposant que les processus en question sont centrés et normalisés convenablement, produit une composante de diffusion à la limite. Finalement, on prouve la récurrence au sens de Harris de X en utilisant un schéma régénératif entièrement basé sur les sauts du processus. De plus, nous dégageons des conditions explicites par rapport aux coefficients du processus qui nous permettent de contrôler la vitesse de convergence vers l'équilibre en terme d'inégalités de déviation pour des fonctionnelles additives intégrables. Dans la seconde partie, nous considérons à nouveau le même type de processus X = (Xt(x)) partant du point x. Utilisant une approche basé sur un Calcul de Malliavin fini-dimensionnel, nous étudions la régularité jointe de ce processus dans le sens suivant : on fixe b≥1 et p>1, K un ensemble compact de Rᵈ, et nous donnons des conditions suffisantes pour avoir P(Xt(x)∈dy)=pt(x,y)dy avec (x,y)↦pt(x,y) appartenant à Wᵇᵖ(K×Rᵈ) / This work is dedicated to the study of some properties concerning the d-dimensional jump type diffusion X = (Xt) with infinitesimal generator given by Lψ(x) = 1/2 ∑ aᵤᵥ(x)∂²ψ(x)/∂xᵤ∂xᵥ + g(x)∇ψ(x) + ∫ (ψ(x + c(z, x)) − ψ(x))γ(z, x)µ(dz) where µ is of infinite total mass. If γ did not depend on x, we would be in a classical situation where the process X could be represented as the solution of a stochastic equation driven by a Poisson point measure with intensity measure γ(z)µ(dz) ; when γ depends on x, we may have the heuristic idea that, if we were to imagine the process as a trajectory of a particle, the law of the jumps may depend on the position of the particle. In the first part, we give some conditions to obtain existence and uniqueness of such processes. Then, we consider this type of processes as a generalization of Piecewise Deterministic Markov Processes (PDMP) ; we show that they can be seen as a limit of a sequence (Xᵣ(t)) of standard PDMP's for which the intensity of the jumps tends to infinity as r tends to infinity, following two regimes: a slow one, which leads to a jump component with finite variation, and a rapid one which, supposing that the processes at hand are centered and renormalized in a convenient way, produces the diffusion component in the limit. Finally, we prove Harris recurrence of X using a regeneration scheme which is entirely based on the jumps of the process. Moreover we state explicit conditions in terms of the coefficients of the process allowing to control the speed of convergence to equilibrium in terms of deviation inequalities for integrable additive functionals. In the second part, we consider again the same type of process X = (Xt(x)) starting from x. Using an approach based on a finite dimensional Malliavin Calculus, we study the joint regularity of this process in the following sense : we fix b≥1 and p>1, K a compact set of Rᵈ, and we give sufficient conditions in order to have P(Xt(x)∈dy)=pt(x,y)dy with (x,y)↦pt(x,y) in Wᵇᵖ(K×Rᵈ)

Robust fault analysis (FA) including the diagnosis of faults and predicting their level of severity is necessary to optimise maintenance and improve reliability of Aircraft. Early diagnosis of faults that might occur in the supervised process renders it possible to perform important preventative actions. The proposed diagnostic models were validated in two experimental tests. The first test concerned a single localised and generalised roller element bearing fault in a permanent magnet brushless DC (PMBLDC) motor. Rolling element bearing defect is one of the main reasons for breakdown in electrical machines. Vibration and current are analysed under stationary and non-stationary load and speed conditions, for a variety of bearing fault severities, and for both local and global bearing faults. The second test examined the case of an unbalance rotor due to blade faults in a thruster, motor based on a permanent magnet brushed DC (PMBDC) motor. A variety of blade fault conditions were investigated, over a wide range of rotation speeds. The test used both discrete wavelet transform (DWT) to extract the useful features, and then feature reduction techniques to avoid redundant features. This reduces computation requirements and the time taken for classification by the application of an orthogonal fuzzy neighbourhood discriminant analysis (OFNDA) approach. The real time monitoring of motor operating conditions is an advanced technique that presents the real performance of the motor, so that the dynamic recurrent neural network (DRNN) proposed predicts the conditions of components and classifies the different faults under different operating conditions. The results obtained from real time simulation demonstrate the effectiveness and reliability of the proposed methodology in accurately classifying faults and predicting levels of fault severity.

Magister Administrationis - MAdmin / The effectiveness of foreign aid in bringing economic and social development is mired in controversy. However, despite the controversial debates on its effectiveness, poor countries of the world have been receiving and using aid as a leverage to relieve themselves from development constraints they faced. Ethiopia is no exception amongst developing countries. Since the time it joined the World Bank group in 1945, foreign capital inflow has remained an important source of revenue for the government. This paper examines the fiscal impact of aid inflow into Ethiopia using time series data for the period 1975-2005. The empirical findings reveal that inflow of foreign aid influences public decision on revenue and expenditure patterns. The result shows that a larger proportion of aid is allocated to capital expenditure and that only a small proportion goes to recurrent expenditure. There is a strong positive association between aid inflow and capital expenditure. The finding also shows that, while a very weak negative association exists between aid and taxation effort, aid and borrowing are used as alternative source of finance. / South Africa

Magister Scientiae (Physiotherapy) - MSc(Physio) / The aim of this study was to establish the factors associated with sports injuries among first division Rwandan female soccer players. Conclusion: Advocacy about the safety and improvement of Rwandan female soccer through workshops and trainings involving players, coaches and team owners need to be emphasised. / South Africa

Ménière’s disease is the third most common inner ear disorder. The individual course of Ménière’s disease in different patients makes it difficult to diagnose on the basis of symptomatology alone. The impact of Ménière’s disease on quality of life has highlighted the importance of an additional tool to support the diagnosis of Ménière’s disease. Apart from the patient’s history, audiological data provide the most relevant information for confirming the diagnosis. The aim of this study was to analyse and describe the clinical and audiological features of a cohort of subjects diagnosed with Ménière’s disease, in order to develop understanding of the pathophysiology of the disease and to facilitate the diagnostic process. The research is based on a retrospective study of the medical records of 135 subjects with Ménière’s disease which were selected according to a non-probability sample. Descriptive statistics were used to organize, analyse and interpret the data. Sixty one percent of subjects presented with definite Ménière’s disease, 14 % with probable Ménière’s disease and 25 % with possible Ménière’s disease. The results showed a higher incidence of Ménière’s disease in females especially in the vestibular type. Three percent of subjects indicated a family history of Ménière’s disease. Bilateral Ménière’s disease presented in 39 % of subjects. The results confirmed that vertigo was the most debilitating symptom in Ménière’s disease. Correlating the clinical features of subjects with audiometric and vestibular tests highlighted the clinical value of an audiological test battery including the following tests: Pure tone audiometry, Speech discrimination, Oto-acoustic emissions, Electronystagmography and Electrocochleography. This confirms the role of the audiologist in the diagnostic and rehabilitation process in patients with Ménière’s disease. / Dissertation (Communication Pathology)--University of Pretoria, 2007. / Speech-Language Pathology and Audiology / unrestricted

Les fièvres récurrentes héréditaires (FRH) sont des maladies auto-inflammatoires transmises selon un mode mendélien. Elles se caractérisent par des accès fébriles récurrents spontanément résolutifs accompagnés d'une inflammation systémique et d'une atteinte des séreuses. La complication la plus grave réside dans le risque de survenue d'une amylose inflammatoire, essentiellement rénale. Le diagnostic clinique des FRH est difficile à établir du fait d'une part d'une grande variabilité inter et intra familiale des phénotypes complexes qui peuvent combiner des signes évocateurs de plusieurs FRH, et d'autre part de l'absence, dans la majorité des cas, de critères objectifs de diagnostic. Alors que le diagnostic de certitude repose essentiellement sur l'identification de défauts moléculaires dans des gènes de l'immunité innée (comme NLRP3, NLRP12, ou MEFV), ces mutations ne rendent compte de la pathologie que chez moins de 30% des cas. Le retentissement fonctionnel de ces variations de séquence, qui sont essentiellement des mutations faux-sens, souvent conservatives, n'a été étudié que dans des lignées cellulaires qui n'expriment pas plusieurs acteurs clés de l'inflammasome, un complexe multiprotéique activé chez les patients présentant une FRH. Au cours de cette thèse, nous avons développé un modèle cellulaire pertinent des FRH à partir de cultures primaires de macrophages humains, dans le but d'étudier les conséquences fonctionnelles des mutations identifiées dans les gènes de FRH et de caractériser les réseaux moléculaires auxquels appartiennent les protéines codées par ces gènes. En parallèle, nous avons cherché à identifier de nouveaux gènes impliqués dans les FRH. / Hereditary recurrent fevers (HRF) define a group of auto-inflammatory diseases transmitted in a Mendelian fashion. They are characterized by recurrent episodes of fever spontaneously resolved, accompanied by systemic inflammation, usually revealed by sterile arthritis, peritonitis, and/or pleurisy. The most serious complication in HRFs is the risk of inflammatory amyloidosis, mainly renal. The clinical diagnosis of HRF is challenging due on the one hand to the inter- and intra- family variability and to complex phenotypes, which combine signs suggestive of different HRFs, and on the other hand, to the absence of objective diagnostic criteria in the majority of cases. While definitive diagnosis is mainly based on the identification of molecular defects in genes of innate immunity (as NLRP3, NLRP12 or MEFV), mutations in these genes account for the pathology in a limited number of patients (30% of cases in our experience). The functional impact of these sequence variations, which are mainly conserved missense mutations, has been studied mainly in heterologous cell lines that do not express several key players of the inflammasome, a multiprotein complex active in patients with HRF. In this thesis, we developed a physiologically relevant cell model of HRF using primary human macrophages in order to assess the functional consequences of the disease-causing mutations and to characterize the molecular networks to which the involved proteins belong. In parallel, we sought to identify novel genes involved in HRF.

O aborto espontâneo recorrente (AER) caracteriza-se pela ocorrência de três ou mais abortos consecutivos espontâneos até a 20ª semana de gestação. É uma condição patológica multifatorial, em que alterações morfológicas uterinas, distúrbios endócrinos, alterações no cariótipo, polimorfismos genéticos relacionados aos genes envolvidos no metabolismo da homocisteína, hemostasia, infecções, autoanticorpos e o processo inflamatório podem contribuir para a ocorrência de AER. O estado fisiológico do endométrio é essencial para a implantação do embrião no útero durante a gestação. Na interface materno-fetal, há uma modulação de citocinas, necessária para o estabelecimento da angiogênese e desenvolvimento da placenta. Um desequilíbrio entre as citocinas pode diminuir a tolerância ao feto e ocasionar rejeição fetal. A concentração de citocinas pode ser modificada por conta de uma diminuição na expressão de alguns genes, e esta pode ser regulada pelo seu estado de metilação sítio-específica. A metilação do DNA é um mecanismo epigenético de regulação gênica, e que corresponde à incorporação de grupos metila em ilhas CpG localizadas próximas às regiões promotoras de genes humanos, e isso pode ser importante na avaliação do risco de complicações gestacionais. Além disso, o estado nutricional de vitaminas foi relacionado a alterações no padrão de metilação de alguns genes. Os objetivos deste trabalho foram avaliar as concentrações dos mediadores inflamatórios em mulheres com aborto e em grupo controle, verificar correlações entre as concentrações de vitaminas, homocisteína total e taxa de metilação do DNA, verificar correlações entre concentração de citocinas e taxa de metilação do DNA e determinar odds ratio (IC 95%) de ter aborto em modelos multivariados. Foram incluídas 253 mulheres com história de aborto recorrente e 264 mulheres saudáveis (controle). O DNA foi extraído de leucócitos de sangue periférico para o estudo de metilação. Foram separadas alíquotas de soro e plasma para dosagem de vitaminas, metabólitos e citocinas. Não foram encontradas diferenças nas taxas de metilação do DNA entre os grupos aborto e controle. A citocina TNFα está aumentada nos grupos de aborto em comparação ao controle. A taxa de metilação do DNA no gene IFNG foi correlacionada inversamente às concentrações de folato sérico e citocina IFNγ no grupo controle. E as concentrações de IL10 foram inversamente correlacionadas à taxa de metilação do DNA nos grupos de aborto secundário e controle. Neste trabalho, verificou-se que as vitaminas e as citocinas influenciam na taxa de metilação do DNA do gene IFNG e a citocina pró-inflamatória TNFα apresenta-se aumentada em mulheres com história de aborto. / Recurrent spontaneous abortion (RSA) is characterized by the occurrence of three or more consecutive spontaneous abortions until the 20th week of gestation. It is a multifactorial pathological condition in which morphological uterine, endocrine disorders, changes in the karyotype genetic polymorphisms related to genes involved in homocysteine metabolism, infection, autoimmunity and inflammatory processes may contribute to the occurrence of RSA. The physiological state of the endometrium is essential for embryo implantation in the uterus during pregnancy. In maternal-fetal interface, there is a modulation of cytokines necessary for the establishment and development of placental angiogenesis. An imbalance between cytokines can decrease tolerance to fetus and cause fetal rejection. Concentration of cytokines may be modified due to a decrease in the expression of genes related to some of these cytokines that can be regulated by DNA methylation, which is an epigenetic mechanism of gene regulation and which corresponds to the incorporation of groups Methyl CpG islands located near the promoter regions of human genes, and this may be important in assessing the risk of pregnancy complications. In addition, the nutritional status of vitamins was associated with changes in the methylation pattern of certain genes. The aims of this study were to determine the concentrations of inflammatory mediators in women with abortion and the control group, examine correlations between concentrations of vitamins, total homocysteine and DNA methylation rate, examine correlations between cytokine concentration and DNA methylation and determine odds ratio (95% CI) of having abortion in multivariate models. We included 253 women with a history of recurrent miscarriage and 264 healthy women (control). DNA was extracted from peripheral blood leukocytes for the study of methylation. Serum and plasma aliquots were used for determination of vitamins, metabolites and cytokines. There were no differences in rates of DNA methylation between control and abortion groups. The cytokine TNFα is increased in abortion groups compared to the control. DNA methylation rate in gene IFNG was inversely correlated with serum folate and serum cytokine IFNγ in the control group. Also IL10 concentrations were inversely correlated to DNA methylation rate in groups of miscarriage and secondary control. In this work, it was found that vitamins and cytokines influence DNA methylation rate in the promoter region and are different in the study and control groups.

abstract: Driving is the coordinated operation of mind and body for movement of a vehicle, such as a car, or a bus. Driving, being considered an everyday activity for many people, still has an issue of safety. Driver distraction is becoming a critical safety problem. Speed, drunk driving as well as distracted driving are the three leading factors in the fatal car crashes. Distraction, which is defined as an excessive workload and limited attention, is the main paradigm that guides this research area. Driver behavior analysis can be used to address the distraction problem and provide an intelligent adaptive agent to work closely with the driver, fay beyond traditional algorithmic computational models. A variety of machine learning approaches has been proposed to estimate or predict drivers’ fatigue level using car data, driver status or a combination of them. Three important features of intelligence and cognition are perception, attention and sensory memory. In this thesis, I focused on memory and attention as essential parts of highly intelligent systems. Without memory, systems will only show limited intelligence since their response would be exclusively based on spontaneous decision without considering the effect of previous events. I proposed a memory-based sequence to predict the driver behavior and distraction level using neural network. The work started with a large-scale experiment to collect data and make an artificial intelligence-friendly dataset. After that, the data was used to train a deep neural network to estimate the driver behavior. With a focus on memory by using Long Short Term Memory (LSTM) network to increase the level of intelligence in two dimensions: Forgiveness of minor glitches, and accumulation of anomalous behavior., I reduced the model error and computational expense by adding attention mechanism on the top of LSTM models. This system can be generalized to build and train highly intelligent agents in other domains. / Dissertation/Thesis / Doctoral Dissertation Computer Engineering 2020

No description available.