Uncertainty is pervasive in risk assessment. In ecotoxicological risk assessments, it arises from such sources as a lack of data, the simplification and abstraction of complex situations, and ambiguities in assessment endpoints (Burgman 2005; Suter 1993). When evaluating and managing risks, uncertainty needs to be explicitly considered in order to avoid erroneous decisions and to be able to make statements about the confidence that we can place in risk estimates. Although informative, previous approaches to dealing with uncertainty in ecotoxicological modelling have been found to be limited, inconsistent and often based on assumptions that may be false (Ferson & Ginzburg 1996; Suter 1998; Suter et al. 2002; van der Hoeven 2004; van Straalen 2002a; Verdonck et al. 2003a). In this thesis a Generalised Linear Modelling approach is proposed as an alternative, congruous framework for the analysis and prediction of a wide range of ecotoxicological effects. This approach was used to investigate the results of toxicity experiments on the effect of 2,4-Dichlorophenoxyacetic Acid (2,4-D) formulations and 4-Chlorophenol (4-CP, an associated breakdown product) on Daphnia carinata. Differences between frequentist Maximum Likelihood (ML) and Bayesian Markov-Chain Monte-Carlo (MCMC) approaches to statistical reasoning and model estimation were also investigated. These approaches are inferentially disparate and place different emphasis on aleatory and epistemic uncertainty (O'Hagan 2004). Bayesian MCMC and Probability Bounds Analysis methods for propagating uncertainty in risk models are also compared for the first time. For simple models, Bayesian and frequentist approaches to Generalised Linear Model (GLM) estimation were found to produce very similar results when non-informative prior distributions were used for the Bayesian models. Potency estimates and regression parameters were found to be similar for identical models, signifying that Bayesian MCMC techniques are at least a suitable and objective replacement for frequentist ML for the analysis of exposureresponse data. Applications of these techniques demonstrated that Amicide formulations of 2,4-D are more toxic to Daphnia than their unformulated, Technical Acid parent. Different results were obtained from Bayesian MCMC and ML methods when more complex models and data structures were considered. In the analysis of 4-CP toxicity, the treatment of 2 different factors as fixed or random in standard and Mixed-Effect models was found to affect variance estimates to the degree that different conclusions would be drawn from the same model, fit to the same data. Associated discrepancies in the treatment of overdispersion between ML and Bayesian MCMC analyses were also found to affect results. Bayesian MCMC techniques were found to be superior to the ML ones employed for the analysis of complex models because they enabled the correct formulation of hierarchical (nested) datastructures within a binomial logistic GLM. Application of these techniques to the analysis of results from 4-CP toxicity testing on two strains of Daphnia carinata found that between-experiment variability was greater than that within-experiments or between-strains. Perhaps surprisingly, this indicated that long-term laboratory culture had not significantly affected the sensitivity of one strain when compared to cultures of another strain that had recently been established from field populations. The results from this analysis highlighted the need for repetition of experiments, proper model formulation in complex analyses and careful consideration of the effects of pooling data on characterising variability and uncertainty. The GLM framework was used to develop three dimensional surface models of the effects of different length pulse exposures, and subsequent delayed toxicity, of 4-CP on Daphnia. These models described the relationship between exposure duration and intensity (concentration) on toxicity, and were constructed for both pulse and delayed effects. Statistical analysis of these models found that significant delayed effects occurred following the full range of pulse exposure durations, and that both exposure duration and intensity interacted significantly and concurrently with the delayed effect. These results indicated that failure to consider delayed toxicity could lead to significant underestimation of the effects of pulse exposure, and therefore increase uncertainty in risk assessments. A number of new approaches to modelling ecotoxicological risk and to propagating uncertainty were also developed and applied in this thesis. In the first of these, a method for describing and propagating uncertainty in conventional Species Sensitivity Distribution (SSD) models was described. This utilised Probability Bounds Analysis to construct a nonparametric 'probability box' on an SSD based on EC05 estimates and their confidence intervals. Predictions from this uncertain SSD and the confidence interval extrapolation methods described by Aldenberg and colleagues (2000; 2002a) were compared. It was found that the extrapolation techniques underestimated the width of uncertainty (confidence) intervals by 63% and the upper bound by 65%, when compared to the Probability Bounds (P3 Bounds) approach, which was based on actual confidence estimates derived from the original data. An alternative approach to formulating ecotoxicological risk modelling was also proposed and was based on a Binomial GLM. In this formulation, the model is first fit to the available data in order to derive mean and uncertainty estimates for the parameters. This 'uncertain' GLM model is then used to predict the risk of effect from possible or observed exposure distributions. This risk is described as a whole distribution, with a central tendency and uncertainty bounds derived from the original data and the exposure distribution (if this is also 'uncertain'). Bayesian and P-Bounds approaches to propagating uncertainty in this model were compared using an example of the risk of exposure to a hypothetical (uncertain) distribution of 4-CP for the two Daphnia strains studied. This comparison found that the Bayesian and P-Bounds approaches produced very similar mean and uncertainty estimates, with the P-bounds intervals always being wider than the Bayesian ones. This difference is due to the different methods for dealing with dependencies between model parameters by the two approaches, and is confirmation that the P-bounds approach is better suited to situations where data and knowledge are scarce. The advantages of the Bayesian risk assessment and uncertainty propagation method developed are that it allows calculation of the likelihood of any effect occurring, not just the (probability)bounds, and that the same software (WinBugs) and model construction may be used to fit regression models and predict risks simultaneously. The GLM risk modelling approaches developed here are able to explain a wide range of response shapes (including hormesis) and underlying (non-normal) distributions, and do not involve expression of the exposure-response as a probability distribution, hence solving a number of problems found with previous formulations of ecotoxicological risk. The approaches developed can also be easily extended to describe communities, include modifying factors, mixed-effects, population growth, carrying capacity and a range of other variables of interest in ecotoxicological risk assessments. While the lack of data on the toxicological effects of chemicals is the most significant source of uncertainty in ecotoxicological risk assessments today, methods such as those described here can assist by quantifying that uncertainty so that it can be communicated to stakeholders and decision makers. As new information becomes available, these techniques can be used to develop more complex models that will help to bridge the gap between the bioassay and the ecosystem.

INTRODUCTION Taenia solium cysticercosis (CC) and taeniosis are zoonotic diseases mainly occurring in developing countries of Asia, Africa and Latin America. Humans are the only definitive host of the parasite harbouring the adult tapeworm in the small intestine (taeniosis). Gravid proglottids containing thousands of eggs are daily shed with the faeces in the environment. The pig intermediate host, through its coprophagic behaviour, may consume contaminated human faeces and consequently, ingest eggs that will further develop into the larval stage of the parasite in the muscles and the brain (cysticerci), causing porcine CC. Humans acquire infection with the adult tapeworm by consumption of undercooked or raw pork infected with cysticerci. Humans can also become infected by the cysts by accidental ingestion of T. solium eggs and develop human CC. In humans, cysts may lodge in muscles, subcutaneous tissues, eyes and/or central nervous system causing ocular CC and neurocysticercosis (NCC) in the two latter cases, respectively. While mainly asymptomatic in pigs, CC and particularly NCC may be responsible of severe health disorders in humans, seizures and epilepsy being the most common ones. In the last decade T. solium became the study target of research groups throughout the world and is now more in the focus of international health and animal health organizations. However, the disease remains neglected and many data gaps still hinder a comprehensive estimation of its burden. Therefore, the general objective of this thesis is the development of tools to estimate the impact of T. solium CC on both public health and veterinary public health in regions where the parasite occurs. For this purpose, four specific objectives are followed focusing on taeniosis, porcine and human CC in different endemic areas of the world. Firstly, a stochastic model is developed allowing the estimation of the monetary and health (disability adjusted life year, DALY) burden of CC in three endemic provinces of Cameroon and the identification of data gaps increasing the uncertainty around the disease burden estimates (Objective 1). Secondly, because there is incomplete knowledge on many CC burden assessment parameters, two protocols are designed to study specific factors that may affect the transmission dynamics of the parasite, and possibly, the disease burden estimation. Because the latter estimation may depend on the age of diseased individuals, a community-based study aims at determining the occurrence of human CC and at studying the age-related infection and transmission patterns of the parasite among a population living in an endemic area of Ecuador (Objective 2). The objective of a second field study conducted in the Democratic Republic of Congo is to determine the occurrence of porcine CC at different levels of the pig trade chain in order to detect any variation in the apparent prevalence figures of the disease (Objective 3). Finally, because none of the available tests for the diagnosis of human CC is a gold standard limiting the estimation of the disease occurrence to the apparent prevalence, and since sensitivity and specificity of the diagnostic tests have only been evaluated in clinical settings, i.e. infected diseased versus uninfected non-diseased individuals, a Bayesian approach is used to estimate the true prevalence of human CC and the performances of available immunological tools for its diagnosis in the framework of an epidemiological study (including uninfected and infected diseased and non-diseased individuals) (Objective 4). RESULTS 1. Development of a stochastic model for the estimation of the economic and health (DALY) burden of T. solium cysticercosis The first experimental section of this thesis aims at developing a stochastic model for the estimation of the monetary and health (DALY) burden of CC, taking 3 endemic provinces of Cameroon as example. Both cost and DALY estimations are applied. All necessary parameters are collected and imported in R software. Different probability distributions are used according to the type of information available for each of the parameters. Monte Carlo simulation techniques allow drawing a set of parameters from the underlying distributions through serial iterations. The number of people with NCC-associated epilepsy represents 1.0% of the local population, whereas the number of pigs diagnosed with CC corresponds to 5.6% of the local pig population. The total annual costs due to T. solium CC in West-Cameroon is estimated at 10,255,202 Euro (95% CR 6,889,04814,754,044), of which 4.7% are due to losses in pig husbandry and 95.3% to direct and indirect expenses associated with human CC. The average number of DALYs lost is 9.0 per thousand persons per year (95% CR 2.820.4; meaning that 9 years of life in perfect health per thousand inhabitants of this area are lost due to CC). This study provides the first estimates of the impact of CC in Cameroon and allows identifying the related economical, clinical and epidemiological necessary parameters and related data gaps. Moreover, the stochastic approach developed here allowed quantifying the uncertainty around the estimates due to the lack of knowledge on this neglected disease. 2. Study on the age-related infection and transmission patterns of human CC in an endemic area The second experimental part of this thesis aims at estimating the apparent prevalence of human CC in an endemic area of Ecuador using both serological antigen and antibody detection methods. Besides a high exposure to the parasite (25% of the sampled population had antibodies directed against T. solium cysticerci) contrasting with a low level of active infection (2.9% had circulating parasite antigens), this study shows that positivity patterns of both diagnostic tests strongly depend on the age of the subjects. The proportion of individuals showing antibodies directed against the larval stage of the parasite significantly increases till the age of 40 to become stable later. Simulation models incorporating insights from cohort study data and basic immunology principles are developed to explain the variations in antibody detection positivity pattern and to reflect the dynamics of exposure to the parasite. The simulations suggest a continual exposure to the parasite through all age categories but a variation of the rate of seroreversion depending on the number of exposures, or in other words, on the immunological status of the individuals. The models also indicate that at least 14% of the population of the community is yearly exposed to the parasite. On the other hand, the proportion of individuals infected by living cysticerci is significantly higher in the elderly. These findings agree with clinical observations in Mexico and Brazil. Immunosenescence could explain such an observation since a weaker immune system in the elderly would facilitate the establishment and maintenance of viable cysticerci in the muscles and in the brain in comparison with fully immunocompetent younger individuals. Our findings indicate that age-related CC infection may influence the CC burden estimates. 3. Occurrence of porcine cysticercosis at different levels of the pig trade flow in a suspected endemic area Based on the use of the enzyme-linked immunosorbent assay for the detection of circulating antigens of T. solium metacestodes (B158/B60 Ag-ELISA), the third experimental section of this thesis reports an apparent prevalence of about 40% of active CC in a pig population in a rural area of Bas-Congo (where pigs are reared) and in a pig population in different markets of Kinshasa (where pork is sold). This work also indicates that the marketing of cysticercotic pig may influence the transmission of the parasite. Indeed, while no significant difference is found in the proportion of active infections between the two sites, the intensity of infection is significantly higher in pigs sampled in the villages in Bas-Congo as compared to those in the markets in Kinshasa. It seems that massively infected animals are excluded at a certain level in the pig trade chain. Preliminary informal surveys on common practices in pig husbandry and pig trading indicate that pig farmers and/or buyers select the low infected animals and exclude those who are positive by tongue inspection at village level. More heavily infected pigs are subsequently used for villagers own consumption or sold at local (clandestine) markets. The influence of the cysticercotic pig marketing should be considered when estimating the animal disease burden. Accurate burden estimates will allow establishing and implementing effective food safety policies and regulations or improving the existent ones in order to control and prevent CC and taeniosis. 4. Estimation of the performances of available immunodiagnostic tools for the diagnosis of human CC in the framework of an epidemiological study The last experimental section of this thesis focuses on the use of Bayesian models to estimate the true prevalence of infection with and exposure to T. solium, and the characteristics of serological tests, namely the enzyme-linked immunosorbent assay for the detection of circulating antigens of the metacestode of T. solium (B158/B60 Ag-ELISA), the enzyme-linked immunoelectrotransfer blot assay (EITB) for the detection of antibodies directed against 7 specific T. solium glycoprotein antigens, and the ELISA for the detection of antibodies directed against crude cyst fluid extracts (Ab-ELISA), used in an epidemiological study setting (including uninfected and infected diseased and non-diseased individuals) in an Ecuadorian endemic area. The EITB and B158/B60 Ag-ELISA sensitivity and specificity estimates in the latter conditions are comparable to the same estimates in clinical settings. However, the present study demonstrates the importance of clearly defining a case when conducting such an exercise. Indeed, the sensitivity and specificity of EITB and B158/B60 Ag-ELISA are high for estimating the true prevalence of exposure to (proportion of individuals who have been in contact with the parasite during the precedent year, with or without development of cysticerci, and with or without (cured individuals) current infection with living cysticerci) and infection with T. solium (proportion of individuals currently infected with living cysticerci), respectively, but drop the other way around. In conclusion, since no gold standard test for human CC diagnosis is available, this study provides a tool for estimation of the true CC prevalence, which is essential for the disease burden assessment. CONCLUSIONS AND PERSPECTIVES Even though the World Health Organization (WHO) already drew the map of the global distribution of T. solium, data remain very limited in several parts of the world. Through a literature review and four experimental sections, this thesis shows that the burden of human and porcine CC may be non negligible and deserves a more global assessment. Moreover, this work allows depicting the disease data gaps more precisely and proposes some tools to deal with the lack of knowledge on this neglected zoonosis. The threat of misusing limited data is a fact and one could reject more global burden estimates based hereon. However, this should not be the main issue. The main concerns should rather focus on how to obtain estimates when only fragmentary data are available. For this purpose, two approaches may be considered: (1) a prospective approach aiming at developing more accurate taeniosis and CC diagnostic tools and at collecting more standardized data from different endemic areas of the world; (2) a retrospective approach aiming at using existing data in innovative ways, such as simulation, Bayesian and stochastic models and expert elicitation, may be a valid alternative. In conclusion, the development of new diagnostic and estimation tools is in progress to compensate for the lack of knowledge on taeniosis/CC and for the diagnostic test and study design limitations (including financial, technical and ethical restrictions). Efforts should be made to increase their availability to developing countries and to standardize study designs in order to produce reliable estimates to assess the global burden of T. solium CC. Once burden estimates are available, the cost-effectiveness of prevention and control programs can be tested and help national and international policy- and decision- makers in setting priorities in public health and veterinary public health policy, services and research. INTRODUCTION La cysticercose (CC) et la taeniose dues à Taenia solium sont des maladies zoonotiques sévissant principalement dans les pays en voie de développement dAsie, dAfrique et dAmérique Latine. LHomme est le seul hôte définitif du parasite, infesté par le ver adulte au niveau de lintestin grêle (taeniose). Des proglottis gravides, contenant des milliers dufs sont quotidiennement évacués avec les matières fécales humaines dans lenvironnement. Le porc, hôte intermédiaire coprophage, peut consommer ces matières fécales contaminées et, par conséquent, ingérer des ufs qui se développeront en stade larvaire au niveau des muscles et du cerveau (les cysticerques), causant ainsi la CC porcine. LHomme, en consommant de la viande de porc infestée (de cysticerques) crue ou mal cuite, sinfeste par le ver adulte bouclant ainsi le cycle. LHomme peut également sinfester par ingestion accidentelle dufs et développer la CC humaine. Chez ce dernier, les cysticerques peuvent se loger dans les muscles, le tissu sous-cutané, les yeux et le système nerveux central, causant, dans ces deux derniers cas, respectivement, la CC oculaire et la neurocysticercose (NCC). Alors que la CC est généralement asymptomatique chez le porc, la CC et tout particulièrement la NCC peuvent être responsables daffections sévères chez lHomme, les convulsions et crises dépilepsie étant les plus fréquentes dentre elles. Durant les dernières décennies, T. solium est devenu le sujet détude principal de plusieurs groupes de recherche à travers le monde ainsi que dorganisations internationales de santé publique et de santé animale. Cependant, le manque de données épidémiologiques et cliniques empêche toujours une estimation précise de son impact réel sur la santé. Cest pourquoi, lobjectif principal de cette thèse est de développer des outils permettant destimer limpact de la CC due à T. solium sur la santé publique et la santé publique vétérinaire. Dans ce but, quatre objectifs spécifiques sont fixés. Premièrement, un modèle stochastique est développé pour estimer limpact économique de la CC due à T. solium et son impact sur la santé (« disability adjusted life year », DALY) dans trois provinces du Cameroun. Cette première étude vise également à identifier les données lacunaires qui augmentent lincertitude autour des estimations (Objectif 1). Sur cette base, deux études épidémiologiques sont conduites dans des communautés endémiques dAfrique et dAmérique Latine, focalisées sur certains facteurs pouvant avoir une influence sur la transmission du parasite, et ainsi sur son impact sur la santé. Puisque lestimation de limpact de la CC sur la santé peut dépendre de lâge des individus malades, une étude de terrain a pour but détudier les relations entre la prévalence apparente de la CC et lâge des individus exposés et infestés dans une zone endémique dEquateur (Objectif 2). Une seconde étude de terrain vise à déterminer loccurrence de la CC porcine à différents niveaux de la filière de production et de commercialisation de la viande de porc dans une zone suspectée endémique de la République Démocratique du Congo (RDC) afin de détecter déventuelles variations de la prévalence apparente de la maladie (Objectif 3). Finalement, vu que la sensibilité et la spécificité des tests pour le diagnostic de la CC humaine ont uniquement été évaluées dans le cadre détudes cliniques, cest-à-dire chez des patients souffrant de CC versus des individus non infestés et asymptomatiques, et vu quaucun de ces tests nest un test parfait (test ayant une sensibilité et une spécificité de 100%), une approche bayésienne est appliquée pour estimer les performances de tests immunologiques disponibles dans le cadre dune étude épidémiologique, incluant des individus non infestés et infestés avec ou sans symptômes (Objectif 4). RESULTATS 1. Développement dun modèle stochastique pour lestimation de limpact économique et de limpact sur la santé (DALY) de la cysticercose due à T. solium Le coût monétaire et le nombre dannées de vie en bonne santé perdues en raison dune mortalité prématurée et/ou de la morbidité dues à la cysticercose (CC) (DALY) sont tous deux estimés dans trois provinces de lOuest du Cameroun. Tous les paramètres épidémiologiques et économiques nécessaires sont collectés et importés dans le logiciel R. Différentes distributions de probabilité sont utilisées en fonction du type dinformation disponible pour chacun dentre eux. Des techniques de simulation de Monte Carlo permettent les estimations finales sur la base des distributions de probabiltés. En tenant compte dune prévalence dépilepsie de 3,6%, le nombre dindividus souffrant dépilepsie associée à la NCC sélève à 1,0% de la population locale. Dautre part, le nombre de porcs diagnostiqués comme infestés correspond à 5,6% de la population porcine locale. Le coût annuel total associé à la CC à lOuest du Cameroun est estimé à 10255202 Euros (95% de la région de confidence (RC) : 6889048-14754044 Euros), desquels 4,7% sont dus aux pertes liées à la production porcine. Le coût par cas de CC humaine séleve à 194 Euros (95% RC : 147-253 Euros). Le nombre moyen de DALYs perdus par 1000 personnes et par an est de 9,0 (95% RC : 2,8 20,4). Cette étude permet une première quantification de limpact de la CC au Cameroun, recense les paramètres économiques, cliniques et épidémiologiques nécessaires à cette fin et le manque de données leur étant lié. De plus, lapproche stochastique utilisée ici permet de quantifier lincertitude autour de ces estimations due au manque de connaissance de cette maladie négligée. 2. Etude de la relation entre la prévalence et la transmission de la cysticercose et lâge des individus exposés et infestés La seconde section expérimentale de cette thèse a pour objectif destimer la prévalence apparente de la CC humaine dans une région endémique de lEquateur en utilisant des techniques sérologiques de détection dantigènes et danticorps. En plus dune exposition élevée au parasite (25% des individus ont des anticorps dirigés contre les cysticerques) contrastant avec une faible proportion dindividus infestés par des cysticerques vivants (3% des individus ont des antigènes circulants de cysticerques), cette étude montre que la positivité aux tests sérologiques dépend de lâge des individus. En effet, la proportion dindividus montrant une réaction humorale dirigée contre le stade larvaire du parasite augmente significativement jusquà lâge de 40 ans pour se stabiliser ensuite. Des modèles de simulation intégrant des données issues dune étude longitudinale ainsi que des principes de bases dimmunologie sont développés dans le but dexpliquer ces variations et de refléter certains aspects de la dynamique dexposition au parasite. Ces simulations suggèrent un taux dexposition au parasite stable à travers les différentes catégories dâge, mais une variation du taux de séroréversion dépendant du nombre de contacts préalables avec le parasite ou, en dautres mots, du statut immunologique des individus. Les modèles indiquent également quau moins 14% de la population sont annuellement exposés au parasite. Dautre part, la proportion des individus infestés par des cysticerques vivants est significativement plus élevée chez les personnes âgées de plus de 60 ans. Limmunosénescence pourrait expliquer ce phénomène, puisque un système immunitaire plus faible chez les personnes âgées pourrait faciliter létablissement et le maintien de cysticerques vivants au sein des muscles et du cerveau. Nos observations indiquent que lâge des individus malades pourrait influencer lestimation de limpact de la CC. 3. Occurrence de la cysticercose porcine à différents niveaux de la filière de la production de viande porcine dans une zone suspectée endémique Sur la base de lutilisation dun test immuno-enzymatique (« enzyme-linked immunosorbent assay », ELISA) pour la détection dantigènes circulants de cysticerques (B158/B60 Ag-ELISA), la troisième section expérimentale de cette thèse permet destimer une prévalence denviron 40% de CC active au sein dune population de porcs élevés traditionnellement dans une zone rurale de la province du Bas-Congo (RDC) et dans une population de porcs vendus et abattus sur différents marchés de Kinshasa (RDC). Ce travail démontre aussi linfluence du commerce de la viande de porc infestée sur la transmission du parasite. En effet, alors quaucune différence significative nest observée entre les proportions de porcs infestés sur les des deux sites détude, lintensité de linfestation est significativement plus élevée chez les porcs des villages que chez les porcs des marchés. Il semble ainsi que les animaux hautement infestés sont exclus à un certain stade de la filière de production et de commercialisation de la viande de porc. Des enquêtes préliminaires réalisées auprès des éleveurs de porcs et des vendeurs de viande de porc indiquent que les éleveurs et/ou les acheteurs de porcs sélectionnent les animaux faiblement infestés et excluent les animaux positifs au langueyage (inspection de la langue) dans les villages. Les porcs hautement infestés sont, par conséquent, consommés par les villageois ou vendus sur des marchés clandestins. Linfluence possible de la filière de la production et de la commercialisation de la viande porcine sur la prévalence apparente de la CC observée dans cette étude devrait être prise en considération lorsque lon évalue limpact de la maladie sur la santé et la production animale. Des estimations précises de cet impact permettront la mise en place de mesures de contrôle et de prévention adéquates de la CC et de la taeniose. 4. Estimation des performances doutils sérologiques pour le diagnostic de la cysticercose humaine dans le cadre dune étude épidémiologique La dernière section expérimentale de cette thèse vise lutilisation de modèles bayésiens pour lestimation de la prévalence réelle dinfestation par et dexposition à T. solium, et des caractéristiques (sensibilité et spécificité) de trois tests sérologiques, à savoir , un ELISA pour la détection dantigènes circulants de cysticerques de T. solium (B158/B60 Ag-ELISA), un autre test immuno-enzymatique (« enzyme-linked immunoelectrotransfer blot assay », EITB) pour la détection danticorps dirigés contre sept glycoprotéines spécifiques à T. solium, et un ELISA pour la détection danticorps dirigés contre le liquide vésiculaire brut de cysticerques (Ab-ELISA), appliqués dans le cadre dune étude épidémiologique (incluant des individus non infestés et infestés avec ou sans symptômes) dans une région endémique de lEquateur. Les performances de lEITB et du B158/B60 Ag-ELISA sont comparables à celles obtenues dans le cadre détudes cliniques. Cependant, cette étude démontre limportance dune définition claire de ce quest « un cas » lorsque lon sadonne à un tel exercice. En effet, la sensibilité et la spécificité de lEITB et du B158/B60 Ag-ELISA sont élevées lorsquil sagit destimer la prévalence dexposition à T. solium (proportion dindividus ayant été en contact avec le parasite durant lannée précédente, avec ou sans établissement de cysticerques et, avec ou sans présence concomitante de cysticerques vivants) et dinfestation par le parasite (proportion dindividus actuellement infestés par des cysticerques vivants), respectivement, mais chutent dans le cas contraire. En conclusion, vu quaucun test parfait nest disponible à ce jour pour le diagnostic de la CC humaine, cette étude propose un outil destimation de la prévalence réelle de la maladie. Cette estimation est essentielle pour lévaluation de limpact de la CC. CONCLUSIONS ET PERSPECTIVES Même si lOrganisation Mondiale de la Santé (OMS) a déjà dressé une carte de la distribution globale de la CC due à T. solium, les données concernant celle-ci restent limitées et incomplètes dans beaucoup de régions endémiques du globe. A travers une étude de la littérature et quatre sections expérimentales, cette thèse montre que limpact de la CC porcine et humaine peut être non négligeable. De plus, ce travail permet de décrire de façon quantitative le manque de connaissance sur la maladie et propose quelques outils permettant dy pallier. Le risque de mal interpréter ou dextrapoler à tort des données incomplètes existe. Il pourrait donc être légitime de rejeter toute estimation plus globale de limpact dune maladie qui serait basée sur ces données. Cependant, le manque de données ne devrait pas signifier « estimation impossible » mais devrait plutôt soulever la question suivante: « Comment pouvons-nous y pallier? ». Pour cela, deux approches sont envisageables : (1) une approche prospective visant à développer des techniques plus performantes pour le diagnostic de la taeniose et de la CC et à permettre la collecte de données standardisées dans différentes zones endémiques du monde ; (2) une approche rétrospective ayant pour but dutiliser les données existantes de manière innovatrice, cest-à-dire en utilisant par exemple des modèles de simulation, des modèles bayésiens ou l« élicitation dexpert ». En conclusion, de nouveaux outils diagnostiques et techniques destimation sont en développement et en cours de validation pour pallier au manque de connaissances, aux limitations des protocoles détudes de recherches scientifiques (incluant les restrictions financières, techniques et éthiques) et/ou aux limitations des tests diagnostiques mais, de considérables efforts devraient être fournis afin de mettre ces outils à la disposition des pays en voie de développement et de standardiser les protocoles de recherche pour permettre une estimation globale de limpact de la CC. Une fois disponible, cette estimation pourra aider les décideurs politiques et les bailleurs de fonds à fixer les priorités en termes de lois, services et recherche scientifiques en santé publique et santé publique vétérinaire.

Made available in DSpace on 2016-06-02T20:04:53Z (GMT). No. of bitstreams: 1 6424.pdf: 5426262 bytes, checksum: 82bb9386f85845b0d3db787265ea8236 (MD5) Previous issue date: 2014-11-21 / The Generalized Normal (GN) and Generalized lognormal (logGN) distributions are flexible for accommodating features present in the data that are not captured by traditional distribution, such as the normal and the lognormal ones, respectively. These distributions are considered to be tools for the reduction of outliers and for the obtention of robust estimates. However, computational problems have always been the major obstacle to obtain the effective use of these distributions. This paper proposes the Bayesian reference analysis methodology to estimate the GN and logGN. The reference prior for a possible order of the model parameters is obtained. It is shown that the reference prior leads to a proper posterior distribution for all the proposed model. The development of Monte Carlo Markov Chain (MCMC) is considered for inference purposes. To detect possible influential observations in the models considered, the Bayesian method of influence analysis on a case based on the Kullback-Leibler divergence is used. In addition, a scale mixture of uniform representation of the GN and logGN distributions are exploited, as an alternative method in order, to allow the development of efficient Gibbs sampling algorithms. Simulation studies were performed to analyze the frequentist properties of the estimation procedures. Real data applications demonstrate the use of the proposed models. / As distribuições normal generalizada (NG) e lognormal generalizada (logNG) são flexíveis por acomodarem características presentes nos dados que não são capturadas por distribuições tradicionais, como a normal e a lognormal, respectivamente. Essas distribuições são consideradas ferramentas para reduzir as observações aberrantes e obter estimativas robustas. Entretanto o maior obstáculo para a utilização eficiente dessas distribuições tem sido os problemas computacionais. Este trabalho propõe a metodologia da análise de referência Bayesiana para estimar os parâmetros dos modelos NG e logNG. A função a priori de referência para uma possível ordem dos parâmetros do modelo é obtida. Mostra-se que a função a priori de referência conduz a uma distribuição a posteriori própria, em todos os modelos propostos. Para fins de inferência, é considerado o desenvolvimento de métodos Monte Carlo em Cadeias de Markov (MCMC). Para detectar possíveis observações influentes nos modelos considerados, é utilizado o método Bayesiano de análise de influência caso a caso, baseado na divergência de Kullback-Leibler. Além disso, uma representação de mistura de escala uniforme para as distribuições NG e logNG é utilizada, como um método alternativo, para permitir o desenvolvimento de algoritmos de amostrador de Gibbs. Estudos de simulação foram desenvolvidos para analisar as propriedades frequentistas dos processos de estimação. Aplicações a conjuntos de dados reais mostraram a aplicabilidade dos modelos propostos.

Made available in DSpace on 2016-06-02T20:05:59Z (GMT). No. of bitstreams: 1 DissANCP.pdf: 696592 bytes, checksum: 1733e6a92a2421365932309fcb98d372 (MD5) Previous issue date: 2007-04-20 / The Item Response Theory (IRT) is a set of mathematical models representing the probability of an individual to take a correct response of an item and its ability. The purpose of our research is to show the models formulated in the IRT under the skew-normal distributions and to develop flexible alternative models. With this goal in mind we introduced the t-skew distributions (Azzalini et al. 1999) and results similar to Bazan s results are obtained. Some applications using Bayesian methods are also considered. / A Teoria de Resposta ao Item (TRI) e um conjunto de modelos matematicos que representam a probabilidade de um indivıduo dar uma resposta certa a um item (questao) como funcao dos parametros do item e da habilidade do indivıduo. O objetivo de nossa pesquisa e apresentar os modelos propostos na TRI normal assimetrica e desenvolver modelos alternativos mais flexıveis. Com esta finalidade em mente, introduzimos a distribuicao t-assimetrica (Azzalini e Capitanio 1999) e obtemos resultados similares aos obtidos por Bazan (2005). Algumas aplicacoes utilizando metodos bayesianos sao consideradas.

Made available in DSpace on 2016-06-02T20:06:06Z (GMT). No. of bitstreams: 1 4329.pdf: 1652803 bytes, checksum: aad7f31afeb38bcd450c35fd9ed227be (MD5) Previous issue date: 2008-09-02 / Universidade Federal de Sao Carlos / The transactions volume increase brought the fraud increase, which result in a annual loss of billions of reais to all .nancial institutions in the world. Therefore, it.s very important the development of detection methods and fraud prevention. The di¢ cult in modeling this kind of data due the fact the data sets are extremely unbalanced. In this work, a bounded logit model will be proposed for fraud detection. It will also be discussed state- dependent sampling and compared with logit and bounded logit model performances. Two applications, one with a simulated data set and another with a real data set, will be presented. The Bayesian approach to these models will also be discussed. The data set analyses will be implemented in SAS and Winbugs software. / O aumento no volume de transações com cartões de crédito trouxe como consequência o aumento do número de fraudes, o que acarreta em uma perda de bilhões de reais anu- almente à todas instituições .nanceiras do mundo. Com isso é muito importante que metodologias de detecção e prevenção à fraude sejam desenvolvidas. A grande di.cul- dade na modelagem deste tipo de dados é que estes são extremamentes desbalanceados. Neste presente trabalho, será proposto o modelo logito limitado na detecção de fraude. Também será discutido as amostras do tipo state-dependent e comparado os desempenhos dos modelos logito e logito limitado. Duas aplicações, uma com um conjunto de dados simulados e outra com um conjunto de dados reais, serão apresentadas. A abordagem bayesiana para estes modelos também será discutida. As análises dos conjuntos de dados serão realizadas nos softwares SAS e Winbugs.

Made available in DSpace on 2016-06-02T20:06:11Z (GMT). No. of bitstreams: 1 DissCFP.pdf: 956101 bytes, checksum: e21a11e1dc4754a5751b0b0840943082 (MD5) Previous issue date: 2006-03-21 / Financiadora de Estudos e Projetos / In this work, we consider the use of Meta-analysis with a Bayesian approach. Meta-analysis is a statistical technique that combines the results of di¤erent independent studies with purpose to find general conclusions. This term was introduced by Glass (1976) and it has been used when the number of studies about some research project is small. Usually, the models for Meta-analysis assume a large number of parameters and the Bayesian approach using MCMC (Markov Chain Monte Carlo) methods is a good alternative to combine information of independent studies, to obtain accutrate inferences about a specified treatment. As illustration, we consider real medical data sets on di¤erent studies, in which, we consider fixed and random e¤ects models. We also assume mixture of normal distributions for the error of the models. Another application is considered with educational data. / Neste trabalho, consideramos o uso de Meta-análise sob um enfoque Bayesiano. Meta-análise é uma técnica estatística que combina resultados de diversos estudos in-dependentes, com o propósito de descrever conclusões gerais. Este termo foi introduzido por Glass (1976) usado quando o número de estudos sobre alguma pesquisa científica é pequeno. Os modelos propostos para Meta-análise usualmente assumem muitos parâmetros e o enfoque Bayesiano com MCMC (Monte Carlo em Cadeias de Markov) é uma alternativa apropriada para combinar informações de estudos independentes. O uso de modelos Bayesianos hierárquicos permite combinações de vários estudos independentes, para a obtenção de inferências precisas sobre um determinado tratamento. Como ilustração numérica consideramos conjuntos de dados médicos de diferentes estudos e, na análise, utilizamos modelos de efeitos fixos e aleatórios e mistura de distribuições normais para o erro do modelo de regressão. Em uma outra aplicação relacionamos Meta-análise e Educação, através do efeito da espectativa do professor associada ao QI dos estudantes.

Cette thèse présente trois contributions indépendantes. La première partie se concentre sur la modélisation de la moyenne conditionnelle des rendements du marché actions : le rendement espéré du marché. Ce dernier est souvent modélisé à l'aide d'un processus AR(1). Cependant, des études montrent que lors de mauvaises périodes économiques la prédictibilité des rendements est plus élevée. Etant donné que le modèle AR(1) exclut par construction cette propriété, nous proposons d'utiliser un modèle CIR. Les implications sont étudiées dans le cadre d'un modèle espace-état bayésien. La deuxième partie est dédiée à la modélisation de la volatilité des actions et des volumes de transaction. La relation entre ces deux quantités a été justifiée par l'hypothèse de mélange de distribution (MDH). Cependant, cette dernière ne capture pas la persistance de la variance, à la différence des spécifications GARCH. Nous proposons un modèle à deux facteurs combinant les deux approches, afin de dissocier les variations de volatilité court terme et long terme. Le modèle révèle plusieurs régularités importantes sur la relation volume-volatilité. La troisième partie s'intéresse à l'analyse des stratégies d'investissement optimales sous contrainte «drawdown ». Le problème étudié est celui de la maximisation d'utilité à horizon fini pour différentes fonctions d'utilité. Nous calculons les stratégies optimales en résolvant numériquement l'équation de Hamilton-Jacobi-Bellman, qui caractérise le principe de programmation dynamique correspondant. En se basant sur un large panel d'expérimentations numériques, nous analysons les divergences des allocations optimales / This PhD thesis presents three independent contributions. The first part is concentrated on the modeling of the conditional mean of stock market returns: the expected market return. The latter is often modeled as an AR(1) process. However, empirical studies have found that during bad times return predictability is higher. Given that the AR(1) model excludes by construction this property, we propose to use instead a CIR model. The implications of this specification are studied within a flexible Bayesian state-space model. The second part is dedicated to the modeling of stocks volatility and trading volume. The empirical relationship between these two quantities has been justified by the Mixture of Distribution Hypothesis (MDH). However, this framework notably fails to capture the obvious persistence in stock variance, unlike GARCH specifications. We propose a two-factor model of volatility combining both approaches, in order to disentangle short-run from long-run volatility variations. The model reveals several important regularities on the volume-volatility relationship. The third part of the thesis is concerned with the analysis of optimal investment strategies under the drawdown constraint. The finite horizon expectation maximization problem is studied for different types of utility functions. We compute the optimal investments strategies, by solving numerically the Hamilton–Jacobi–Bellman equation, that characterizes the dynamic programming principle related to the stochastic control problem. Based on a large panel of numerical experiments, we analyze the divergences of optimal allocation programs

The Autonomic Power System (APS) grand challenge was a multi-disciplinary EPSRC-funded research project that examined novel techniques that would enable the transition between today's and 2050's highly uncertain and complex energy network. Being part of the APS, this thesis reports on the sub-project 'RR2: Avoiding High-Impact Low Probability events'. The goal of RR2 is to develop new algorithms for controlling risk exposure to high-impact low probability (Hi-Lo) events through the provision of appropriate risk-sensitive control strategies. Additionally, RR2 is concerned with new techniques for identifying and modelling risk in future energy networks, in particular, the risk of Hi-Lo events. In this context, this thesis investigates two distinct problems arising from energy risk management. On the one hand, we examine the problem of finding managerial strategies for exercising the operational flexibility of energy assets. We look at this problem from a risk perspective taking into account non-linear risk preferences of energy asset managers. Our main contribution is the development of a risk-sensitive approach to the class of optimal switching problems. By recasting the problem as an iterative optimal stopping problem, we are able to characterise the optimal risk-sensitive switching strategies. As byproduct, we obtain a multiplicative dynamic programming equation for the value function, upon which we propose a numerical algorithm based on least squares Monte Carlo regression. On the other hand, we develop tools to identify and model the risk factors faced by energy asset managers. For this, we consider a class of models consisting of superposition of Gaussian and non-Gaussian Ornstein-Uhlenbeck processes. Our main contribution is the development of a Bayesian methodology based on Markov chain Monte Carlo (MCMC) algorithms to make inference into this class of models. On extensive simulations, we demonstrate the robustness and efficiency of the algorithms to different data features. Furthermore, we construct a diagnostic tool based on Bayesian p-values to check goodness-of-fit of the models on a Bayesian framework. We apply this tool to MCMC results from fitting historical electricity and gas spot price data- sets corresponding to the UK and German energy markets. Our analysis demonstrates that the MCMC-estimated models are able to capture not only long- and short-lived positive price spikes, but also short-lived negative price spikes which are typical of UK gas prices and German electricity prices. Combining together the solutions to the two problems above, we strive to capture the interplay between risk, uncertainty, flexibility and performance in various applications to energy systems. In these applications, which include power stations, energy storage and district energy systems, we consistently show that our risk management methodology offers a tradeoff between maximising average performance and minimising risk, while accounting for the jump dynamics of energy prices. Moreover, the tradeoff is achieved in such way that the benefits in terms of risk reduction outweigh the loss in average performance.

La modélisation par équations structurelles est très répandue dans des domaines très variés et nous l'appliquons pour la première fois en métrologie dans le traitement de données de comparaisons interlaboratoires. Les modèles à équations structurelles à variables latentes sont des modèles multivariés utilisés pour modéliser des relations de causalité entre des variables observées (les données). Le modèle s'applique dans le cas où les données peuvent être regroupées dans des blocs disjoints où chaque bloc définit un concept modélisé par une variable latente. La structure de corrélation des variables observées est ainsi résumée dans la structure de corrélation des variables latentes. Nous proposons une approche bayésienne des modèles à équations structurelles centrée sur l'analyse de la matrice de corrélation des variables latentes. Nous appliquons une expansion paramétrique à la matrice de corrélation des variables latentes afin de surmonter l'indétermination de l'échelle des variables latentes et d'améliorer la convergence de l'algorithme de Gibbs utilisé. La puissance de l'approche structurelle nous permet de proposer une modélisation riche et flexible des biais de mesure qui vient enrichir le calcul de la valeur de consensus et de son incertitude associée dans un cadre entièrement bayésien. Sous certaines hypothèses l'approche permet de manière innovante de calculer les contributions des variables de biais au biais des laboratoires. Plus généralement nous proposons un cadre bayésien pour l'amélioration de la qualité des mesures. Nous illustrons et montrons l'intérêt d'une modélisation structurelle des biais de mesure sur des comparaisons interlaboratoires en environnement. / Structural equation modelling is a widespread approach in a variety of domains and is first applied here to interlaboratory comparisons in metrology. Structural Equation Models with latent variables (SEM) are multivariate models used to model causality relationships in observed variables (the data). It is assumed that data can be grouped into separate blocks each describing a latent concept modelled by a latent variable. The correlation structure of the observed variables is transferred into the correlation structure of the latent variables. A Bayesian approach of SEM is proposed based on the analysis of the correlation matrix of latent variables using parameter expansion to overcome identifiability issues and improving the convergence of the Gibbs sampler. SEM is used as a powerful and flexible tool to model measurement bias with the aim of improving the reliability of the consensus value and its associated uncertainty in a fully Bayesian framework. The approach also allows to compute the contributions of the observed variables to the bias of the laboratories, under additional hypotheses. More generally a global Bayesian framework is proposed to improve the quality of measurements. The approach is illustrated on the structural equation modelling of measurement bias in interlaboratory comparisons in environment.

Made available in DSpace on 2016-06-02T20:04:51Z (GMT). No. of bitstreams: 1 3028.pdf: 7339557 bytes, checksum: 16711c2271b754604bfa0b0fba30290b (MD5) Previous issue date: 2010-03-04 / In this work, we introduce a Bayesian Analysis for survival multivariate data in the presence of a covariate vector and censored observations. Different frailties or latent variables are considered to capture the correlation among the survival times for the same individual. We also introduce a Bayesian analysis for some of the most popular bivariate exponential distributions introduced in the literature. A Bayesian analysis is also introduced for the Block & Basu bivariate exponential distribution using Markov Chain Monte Carlo (MCMC) methods and considering lifetimes in presence of covariates and censored data. In another topic, we introduce a Bayesian Analysis for bivariate lifetime data in the presence of covariates and censoring data assuming different bivariate Weibull distributions derived from some existing copula functions. A great computational simplification to simulate samples for the joint posterior distribution is obtained using the WinBUGS software. Numerical illustrations are introduced considering real data sets considering every proposed methodology. / Nesta tese introduzimos uma an´alise Bayesiana para dados de sobreviv encia multivariados, na presen¸ca de um vetor de covari´aveis e observa¸c oes censuradas. Diferentes fragilidades ou vari´aveis latentes s ao consideradas para capturar a correla¸c ao existente entre os tempos de sobreviv encia, para o mesmo indiv´ıduo. Tamb´em apresentamos uma an´alise Bayesiana para algumas das mais populares distribui¸c oes exponenciais bivariadas introduzidas na literatura. Uma an´alise Bayesiana tamb´em ´e introduzida para a distribui¸c ao exponencial bivariada de Block & Basu, usando m´etodos MCMC (Monte Carlo em Cadeias de Markov) e considerando os tempos de sobreviv encia na presen¸ca de covari´aveis e dados censurados. Em outro t´opico, introduzimos uma an´alise Bayesiana para dados de sobreviv encia bivariados na presen¸ca de covari´aveis e observa¸c oes censuradas, assumindo diferentes distribui¸c oes bivariadas Weibull derivadas de algumas fun¸c oes c´opulas existentes. Uma grande simplifica¸c ao computacional para simular amostras da distribui¸c ao a posteriori conjunta de interesse ´e obtida usando o software WinBUGS. Ilustra¸c oes num´ericas s ao introduzidas considerando conjunto de dados reais, para cada uma das metodologias propostas.