Le diabète de type 2 (DT2) est une maladie chronique associée à de graves et coûteuses complications. Dans un contexte de restriction budgétaire, il est nécessaire de pouvoir estimer les ressources à affecter à la prise en charge des maladies chroniques et donc de suivre l’évolution épidémiologique et économique d’une telle maladie. Une base de données a été construite à partir des données médico-administratives de l’assurance maladie luxembourgeoise. Elle comprenait les consommations de soins, associées au diabète et ses complications, des patients diabétiques de type 2 traités entre 2000 et 2006. L’objectif était d’étudier les champs d’utilisation de ces données et leurs applications possibles pour les décisions en santé publique. Cette thèse en donne quelques exemples. En 2006, la prévalence du DT2 au Luxembourg était de 3,79% (N= 17 070). Un algorithme a permis d’identifier trois stades de la néphropathie diabétique (3,77% des cas de DT2 en 2006). L’analyse de l’adhérence aux recommandations européennes de bonnes pratiques médicales a mis en évidence une situation critique associée à certains facteurs (médecin traitant, type de traitement, région de résidence…). Les dépenses moyennes d’un patient en hémodialyse a été estimé à 116 647€/patient en 2006. Enfin, une analyse médico-économique a montré la dominance coût-efficace d’une stratégie d’implémentation de la dialyse péritonéale sur la situation actuelle. Malgré les difficultés à évaluer leur qualité, les données médico-administratives offrent une source d’informations précieuses pour les décideurs publics et les professionnels de la santé, dans le but d’améliorer la prise en charge des patients. / Type 2 diabetes (T2D) is a chronic disease associated with many severe and costly complications. In a context of budgetary constraint, it is necessary to obtain an estimate the amount of resources to allocate to the management of chronic diseases. This includes monitoring the epidemiologic and economic evolutions. A database was built from medico-administrative databases of the national health insurance of Luxembourg. It included the healthcare consumptions associated with diabetes and its complications, of all type 2 diabetic patients treated in Luxembourg between 2000 and 2006. The objectives were to study the fields of use of this database and the possible applications for public health decision-making. This thesis gives some examples. In 2006, T2D prevalence in Luxembourg was 3.79% (N= 17070). An algorithm was built and permitted to identify three stages of diabetic nephropathy (3.77% of T2D cases in 2006). The analysis of the adherence to European follow-up guidelines showed a critical situation associated to several factors (treating physician, type of treatment, living region…). The mean costs associated with patients in dialysis were estimated at 116 647€/patient in 2006. Finally, a health-economic evaluation showed the dominance of a strategy promoting peritoneal dialysis in Luxembourg over the present situation.

O desafio para a área de monitoramento de processos, em indústrias químicas, ainda é a etapa de detecção, com a necessidade de desenvolvimento de sistemas confiáveis. Pode-se resumir que um sistema é confiável, ao ser capaz de detectar as situações anormais, de modo precoce, e, ao mesmo tempo, de minimizar a geração de alarmes falsos. Ao se ter um sistema confiável, pode-se empregá-lo para auxiliar o operador, de fábricas, no processo de tomada de decisões. O objetivo deste estudo é apresentar uma metodologia, baseada na técnica, modelo oculto de Markov (HMM, acrônimo de ?Hidden Markov Model?), para se detectar situações anormais em caldeiras de recuperação química. As aplicações de maior sucesso de HMM são na área de reconhecimento de fala. Pode-se citar como aspectos positivos: o raciocínio probabilístico, a modelagem explícita, e a identificação a partir de dados históricos. Fez-se duas aplicações. O primeiro estudo de caso é no ?benchmark? de um sistema de evaporação múltiplo efeito de uma fábrica de produção de açúcar. Identificou-se um HMM, característico de operação normal, para se detectar cinco situações anormais no atuador responsável por regular o fluxo de xarope de açúcar para o primeiro evaporador. A detecção, para as três situações abruptas, é imediata, uma vez que o HMM foi capaz de detectar alterações, abruptas, no sinal da variável monitorada. Em relação às duas situações incipientes, foi possível detectá-las ainda em estágio inicial; ao ser o valor de f (vetor responsável por representar a intensidade de um evento anormal, com o tempo), no instante da detecção, próximo a zero, igual a 2,8% e 2,1%, respectivamente. O segundo estudo de caso é em uma caldeira de recuperação química, de uma fábrica de produção de celulose, no Brasil. O objetivo é monitorar o acúmulo de depósitos de cinzas sobre os equipamentos da sessão de transferência de calor convectivo, através de medições de perda de carga. Este é um dos principais desafios para se aumentar a eficiência operacional deste equipamento. Após a identificação de um HMM característico de perda de carga alta, pôde-se verificar a sua capacidade de informar o estado atual e, por consequência, a tendência do sistema, de modo similar à um preditor. Pôde-se demonstrar também a utilidade de se definir limites de controle, com o objetivo de se ter a informação sobre a distância entre o estado atual e os níveis de alarme de perda de carga. / The greatest challenge faced by the area of process monitoring in chemical industries still resides in the fault detection task, which aims at developing reliable systems. One may say that a system is reliable if it is able to perform early fault detection and, at the same time, to reduce the generation of false alarms. Once there is a reliable system available, it can be employed to help operators, in factories, in the decisionmaking process. The aim of this study is presenting a methodology, based on the Hidden Markov Model (HMM) technique, suggesting its use in the detection of abnormal situations in chemical recovery boilers. The most successful applications of HMM are in the area of speech recognition. Some of its advantages are: probabilistic reasoning, explicit modeling and the identification based on process history data. This study discusses two applications. The first one is on a benchmark of a multiple evaporation system in a sugar factory. A HMM representative of the normal operation was identified, in order to detect five abnormal situations at the actuator responsible for controlling the syrup flow to the first evaporator. The detection result for the three abrupt situations was immediate, since the HMM was capable of detecting the statistical changes on the signal of the monitored variable as soon as they occurred. Regarding to the two incipient situations, the detection was done at an early stage. For both events, the value of vector f (responsible for representing the strength of an abnormal event over time), at the time it occurred, was near zero, equal to 2.8 and 2.1%, respectively. The second case study deals with the application of HMM in a chemical recovery boiler, belonging to a cellulose mill, in Brazil. The aim is monitoring the accumulation of ash deposits over the equipments of the convective heat transfer section, through pressure drop measures. This is one of the main challenges to be overcome nowadays, bearing in mind the interest that exists in increasing the operational efficiency of this equipment. Initially, a HMM for high values of pressure drop was identified. With this model, it was possible to check its capacity to inform the current state, and consequently, the tendency of the system (similarly as a predictor). It was also possible to show the utility of defining control limits, in order to inform the operator the relative distance between the current state of the system and the alarm levels of pressure drop.

No description available.

The growth of multimedia content in recent years motivated the research on audio classification and content retrieval area. In this thesis, a general environmental audio classification and retrieval approach is proposed in which higher level semantic classes (outdoor, nature, meeting and violence) are obtained from lower level acoustic classes (emergency alarm, car horn, gun-shot, explosion, automobile, motorcycle, helicopter, wind, water, rain, applause, crowd and laughter). In order to classify an audio sample into acoustic classes, MPEG-7 audio features, Mel Frequency Cepstral Coefficients (MFCC) feature and Zero Crossing Rate (ZCR) feature are used with Hidden Markov Model (HMM) and Support Vector Machine (SVM) classifiers. Additionally, a new classification method is proposed using Genetic Algorithm (GA) for classification of semantic classes. Query by Example (QBE) and keyword-based query capabilities are implemented for content retrieval.

No description available.

Darbas „Lietuvių šnekos atpažinimo akustinis modeliavimas“ yra skirtas lietuvių šnekos atpažinimo akustiniam modeliavimui. Darbe buvo tirtas žodžiais, skiemenimis, kontekstiniais skiemenimis, fonemomis ir kontekstinėmis fonemomis grįstas šnekos atpažinimas. Tyrimai atlikti izoliuotiems žodžiams ir ištisinei šnekai. Iki šiol lietuvių šnekos atpažinime populiariausi kalbos vienetai buvo fonema ir kontekstinė fonema, o kitų kalbos vienetų analizė nebuvo atliekama. Šiame darbe siekiama palyginti lingvistinio tipo kalbos vienetų gebėjimą modeliuoti šneką ir parodyti, kad kalbos vienetų analizė siūlo alternatyvius fonemai ir kontekstinei fonemai kalbos vienetus. Darbe pasiūlyta metodika mišriam skiemenų ir fonemų akustiniam modeliavimui, naujas kalbos vienetas – pseudo-skiemuo; technologijos atskirų kalbos vienetų akustiniam modeliavimui (schemos, įrankiai, rekomendacijos). Eksperimentiniams tyrimams atlikti paruoštas izoliuotų žodžių garsynas ir sukurtos dvi ištisinės šnekos garsyno LRN versijos. Ištyrus izoliuotų žodžių atpažinimą, akustinius modelius konstruojant žodžiams, nustatyta, kad modelių mokymo aibės dydis, akustinių modelių mokymo aibės turinys daro įtaką šnekos atpažinimo tikslumui. Pateikiamos rekomendacijos akustiniam modeliavimui žodžių pagrindu. Ištyrus izoliuotų žodžių atpažinimą, akustinius modelius konstruojant žodžiams, skiemenims ir fonemoms, gauti rezultatai 98 ±1,8 % tikslumu siejami su skiemens tipo kalbos vienetais. Dėl skiemenų akustinio modeliavimo... [toliau žr. visą tekstą] / This paper is devoted to an acoustic modelling of Lithuanian speech recognition. Word-, syllable-, contextual syllable-, phoneme- and contextual phoneme-based speech recognition was investigated. Investigations were performed for isolated words and continuous speech. The most popular sub-word units in Lithuanian speech recognition are phonemes and contextual phonemes, and research on other sub-word units is omitted. This paper aims to compare capacity of linguistic sub-word units to model speech and to demonstrate that investigation of sub-word units suggest using alternative sub-word units to phoneme and contextual phoneme. The dissertation proposes a new methodology for acoustic modelling of syllables and phonemes, new sub-word unit – pseudo-syllable; technologies for acoustic modelling of separate sub-word units, including developed schemes, tools and recommendations. Speech corpus of isolated words was prepared and two versions of corpus of continuous speech LRN were developed for experimental research. Investigation of recognition of isolated words and construction of acoustic models for words showed that a size of training set of acoustic models, a content of training set in regard to number of speakers have an influence on speech recognition accuracy. The recommendations for word-based acoustic modelling are given. Investigation of recognition of isolated words and construction of acoustic models for words, syllables and phonemes showed that the best recognition... [to full text]

This paper is devoted to an acoustic modelling of Lithuanian speech recognition. Word-, syllable-, contextual syllable-, phoneme- and contextual phoneme-based speech recognition was investigated. Investigations were performed for isolated words and continuous speech. The most popular sub-word units in Lithuanian speech recognition are phonemes and contextual phonemes, and research on other sub-word units is omitted. This paper aims to compare capacity of linguistic sub-word units to model speech and to demonstrate that investigation of sub-word units suggest using alternative sub-word units to phoneme and contextual phoneme. The dissertation proposes a new methodology for acoustic modelling of syllables and phonemes, new sub-word unit – pseudo-syllable; technologies for acoustic modelling of separate sub-word units, including developed schemes, tools and recommendations. Speech corpus of isolated words was prepared and two versions of corpus of continuous speech LRN were developed for experimental research. Investigation of recognition of isolated words and construction of acoustic models for words showed that a size of training set of acoustic models, a content of training set in regard to number of speakers have an influence on speech recognition accuracy. The recommendations for word-based acoustic modelling are given. Investigation of recognition of isolated words and construction of acoustic models for words, syllables and phonemes showed that the best recognition... [to full text] / Darbas „Lietuvių šnekos atpažinimo akustinis modeliavimas“ yra skirtas lietuvių šnekos atpažinimo akustiniam modeliavimui. Darbe buvo tirtas žodžiais, skiemenimis, kontekstiniais skiemenimis, fonemomis ir kontekstinėmis fonemomis grįstas šnekos atpažinimas. Tyrimai atlikti izoliuotiems žodžiams ir ištisinei šnekai. Iki šiol lietuvių šnekos atpažinime populiariausi kalbos vienetai buvo fonema ir kontekstinė fonema, o kitų kalbos vienetų analizė nebuvo atliekama. Šiame darbe siekiama palyginti lingvistinio tipo kalbos vienetų gebėjimą modeliuoti šneką ir parodyti, kad kalbos vienetų analizė siūlo alternatyvius fonemai ir kontekstinei fonemai kalbos vienetus. Darbe pasiūlyta metodika mišriam skiemenų ir fonemų akustiniam modeliavimui, naujas kalbos vienetas – pseudo-skiemuo; technologijos atskirų kalbos vienetų akustiniam modeliavimui (schemos, įrankiai, rekomendacijos). Eksperimentiniams tyrimams atlikti paruoštas izoliuotų žodžių garsynas ir sukurtos dvi ištisinės šnekos garsyno LRN versijos. Ištyrus izoliuotų žodžių atpažinimą, akustinius modelius konstruojant žodžiams, nustatyta, kad modelių mokymo aibės dydis, akustinių modelių mokymo aibės turinys daro įtaką šnekos atpažinimo tikslumui. Pateikiamos rekomendacijos akustiniam modeliavimui žodžių pagrindu. Ištyrus izoliuotų žodžių atpažinimą, akustinius modelius konstruojant žodžiams, skiemenims ir fonemoms, gauti rezultatai 98 ±1,8 % tikslumu siejami su skiemens tipo kalbos vienetais. Dėl skiemenų akustinio modeliavimo... [toliau žr. visą tekstą]

Advances in genomics allow researchers to measure the complete set of transcripts in cells. These transcripts include messenger RNAs (which encode for proteins) and microRNAs, short RNAs that play an important regulatory role in cellular networks. While this data is a great resource for reconstructing the activity of networks in cells, it also presents several computational challenges. These challenges include the data collection stage which often results in incomplete and noisy measurement, developing methods to integrate several experiments within and across species, and designing methods that can use this data to map the interactions and networks that are activated in specific conditions. Novel and efficient algorithms are required to successfully address these challenges. In this thesis, we present probabilistic models to address the set of challenges associated with expression data. First, we present a novel probabilistic error correction method for RNA-Seq reads. RNA-Seq generates large and comprehensive datasets that have revolutionized our ability to accurately recover the set of transcripts in cells. However, sequencing reads inevitably contain errors, which affect all downstream analyses. To address these problems, we develop an efficient hidden Markov modelbased error correction method for RNA-Seq data . Second, for the analysis of expression data across species, we develop clustering and distance function learning methods for querying large expression databases. The methods use a Dirichlet Process Mixture Model with latent matchings and infer soft assignments between genes in two species to allow comparison and clustering across species. Third, we introduce new probabilistic models to integrate expression and interaction data in order to predict targets and networks regulated by microRNAs. Combined, the methods developed in this thesis provide a solution to the pipeline of expression analysis used by experimentalists when performing expression experiments.

Geodetic adjustment models are often set up in a way that the model parameters need to fulfil certain constraints. The normalized Lagrange multipliers have been used as a measure of the strength of constraint in such a way that if one of them exceeds in magnitude a certain threshold then the corresponding constraint is likely to be incompatible with the observations and the rest of the constraints. We show that these and similar measures can be deduced as test statistics of a likelihood ratio test of the statistical hypothesis that some constraints are incompatible in the same sense. This has been done before only for special constraints (Teunissen in Optimization and Design of Geodetic Networks, pp. 526–547, 1985). We start from the simplest case, that the full set of constraints is to be tested, and arrive at the advanced case, that each constraint is to be tested individually. Every test is worked out both for a known as well as for an unknown prior variance factor. The corresponding distributions under null and alternative hypotheses are derived. The theory is illustrated by the example of a double levelled line. / Geodätische Ausgleichungsmodelle werden oft auf eine Weise formuliert, bei der die Modellparameter bestimmte Bedingungsgleichungen zu erfüllen haben. Die normierten Lagrange-Multiplikatoren wurden bisher als Maß für den ausgeübten Zwang verwendet, und zwar so, dass wenn einer von ihnen betragsmäßig eine bestimmte Schwelle übersteigt, dann ist davon auszugehen, dass die zugehörige Bedingungsgleichung nicht mit den Beobachtungen und den restlichen Bedingungsgleichungen kompatibel ist. Wir zeigen, dass diese und ähnliche Maße als Teststatistiken eines Likelihood-Quotiententests der statistischen Hypothese, dass einige Bedingungsgleichungen in diesem Sinne inkompatibel sind, abgeleitet werden können. Das wurde bisher nur für spezielle Bedingungsgleichungen getan (Teunissen in Optimization and Design of Geodetic Networks, pp. 526–547, 1985). Wir starten vom einfachsten Fall, dass die gesamte Menge der Bedingungsgleichungen getestet werden muss, und gelangen zu dem fortgeschrittenen Problem, dass jede Bedingungsgleichung individuell zu testen ist. Jeder Test wird sowohl für bekannte, wie auch für unbekannte a priori Varianzfaktoren ausgearbeitet. Die zugehörigen Verteilungen werden sowohl unter der Null- wie auch unter der Alternativhypthese abgeleitet. Die Theorie wird am Beispiel einer Doppelnivellementlinie illustriert.

Many complexity measures are defined as the size of a minimal representation in a specific model class. One such complexity measure, which is important because it is widely applied, is statistical complexity. It is defined for discrete-time, stationary stochastic processes within a theory called computational mechanics. Here, a mathematically rigorous, more general version of this theory is presented, and abstract properties of statistical complexity as a function on the space of processes are investigated. In particular, weak-* lower semi-continuity and concavity are shown, and it is argued that these properties should be shared by all sensible complexity measures. Furthermore, a formula for the ergodic decomposition is obtained. The same results are also proven for two other complexity measures that are defined by different model classes, namely process dimension and generative complexity. These two quantities, and also the information theoretic complexity measure called excess entropy, are related to statistical complexity, and this relation is discussed here. It is also shown that computational mechanics can be reformulated in terms of Frank Knight's prediction process, which is of both conceptual and technical interest. In particular, it allows for a unified treatment of different processes and facilitates topological considerations. Continuity of the Markov transition kernel of a discrete version of the prediction process is obtained as a new result.