La plupart des patients infectés par le VIH ont une charge virale qui peut être rendue indétectable par des combinaisons antirétrovirales hautement actives (cART); cependant, il existe des effets secondaires aux traitements. L'utilisation des modèles mécanistes dynamiques basés sur des équations différentielles ordinaires (ODE) a considérablement amélioré les connaissances de la dynamique HIV-système immunitaire et permet d'envisager une personnalisation du traitement. L'objectif de ces travaux de thèse est d'améliorer les techniques statistiques d'estimation de paramètres dans les modèles mécanistes dynamiques afin de proposer des stratégies de surveillance et d'optimisation des traitements. Après avoir introduit NIMROD un algorithme d'estimation bayésienne basé sur une maximisation de la vraisemblance pénalisée, nous montrons la puissance des approches mécanistes dynamiques pour l'évaluation des effets traitements par rapport aux méthodes descriptives d'analyse des trajectoires des biomarqueurs. Puis, nous définissons le « modèle à cellules cibles », un système ODE décrivant la dynamique du VIH et des CD4. Nous montrons qu'il possède de bonnes capacités prédictives. Nous proposons une preuve de concept de la possibilité de contrôler individuellement la dose de traitement. Cette stratégie adaptative réajuste la dose du patient en fonction de sa réaction à la dose précédente par une procédure bayésienne. Pour finir, nous introduisons la possibilité de l’'individualisation des changements de cART. Ce travail passe par la quantification in vivo d'effets de cART en utilisant des indicateurs d'activité antivirale in vitro. Nous discutons la validité des résultats et les étapes méthodologiques nécessaires pour l'intégration de ces méthodes dans les pratiques cliniques. / Most HIV-infected patients viral loads can be made undetectable by highly active combination of antiretroviral therapy (cART), but there are side effects of treatments. The use of dynamic mechanistic models based on ordinary differential equations (ODE) has greatly improved the knowledge of the dynamics of HIV and of the immune system and can be considered for personalization of treatment. The aim of these PhD works is to improve the statistical techniques for estimating parameters in dynamic mechanistic models so as to elaborate strategies for monitoring and optimizing treatments. We present an algorithm and program called NIMROD using Bayesian inference based on the maximization of the penalized likelihood. Then, we show the power of dynamic mechanistic approaches for the evaluation of treatment effects compared to methods based on the descriptive analysis of the biomarkers trajectories. Next, we build the “target cells model “, an ODE system of the dynamics between the HIV and CD4. We demonstrate it has good predictive capabilities. We build a proof of concept for drug dose individualization. It consists in tuning the dose of the patient based on his reaction to the previous doses using a Bayesian update procedure. Finally, we introduce the possibility of designing an individualized change of cART. This work involves the quantification of in vivo effects of cART using in vitro antiviral activity indicators. We discuss the validity of the results and the further steps needed for the integration of these methods in clinical practice.

Series data, sequences of measured values, are ubiquitous. Whenever observations are made along a path in space or time, a data sequence results. To comprehend nature and shape it to our will, or to make informed decisions based on what we know, we need methods to make sense of such data. Of particular interest are probabilistic descriptions, which enable us to represent uncertainty and random variation inherent to the world around us. This thesis presents and expands upon some tools for creating probabilistic models of sequences, with an eye towards applications involving speech and language. Modelling speech and language is not only of use for creating listening, reading, talking, and writing machines---for instance allowing human-friendly interfaces to future computational intelligences and smart devices of today---but probabilistic models may also ultimately tell us something about ourselves and the world we occupy. The central theme of the thesis is the creation of new or improved models more appropriate for our intended applications, by weakening limiting and questionable assumptions made by standard modelling techniques. One contribution of this thesis examines causal-state splitting reconstruction (CSSR), an algorithm for learning discrete-valued sequence models whose states are minimal sufficient statistics for prediction. Unlike many traditional techniques, CSSR does not require the number of process states to be specified a priori, but builds a pattern vocabulary from data alone, making it applicable for language acquisition and the identification of stochastic grammars. A paper in the thesis shows that CSSR handles noise and errors expected in natural data poorly, but that the learner can be extended in a simple manner to yield more robust and stable results also in the presence of corruptions. Even when the complexities of language are put aside, challenges remain. The seemingly simple task of accurately describing human speech signals, so that natural synthetic speech can be generated, has proved difficult, as humans are highly attuned to what speech should sound like. Two papers in the thesis therefore study nonparametric techniques suitable for improved acoustic modelling of speech for synthesis applications. Each of the two papers targets a known-incorrect assumption of established methods, based on the hypothesis that nonparametric techniques can better represent and recreate essential characteristics of natural speech. In the first paper of the pair, Gaussian process dynamical models (GPDMs), nonlinear, continuous state-space dynamical models based on Gaussian processes, are shown to better replicate voiced speech, without traditional dynamical features or assumptions that cepstral parameters follow linear autoregressive processes. Additional dimensions of the state-space are able to represent other salient signal aspects such as prosodic variation. The second paper, meanwhile, introduces KDE-HMMs, asymptotically-consistent Markov models for continuous-valued data based on kernel density estimation, that additionally have been extended with a fixed-cardinality discrete hidden state. This construction is shown to provide improved probabilistic descriptions of nonlinear time series, compared to reference models from different paradigms. The hidden state can be used to control process output, making KDE-HMMs compelling as a probabilistic alternative to hybrid speech-synthesis approaches. A final paper of the thesis discusses how models can be improved even when one is restricted to a fundamentally imperfect model class. Minimum entropy rate simplification (MERS), an information-theoretic scheme for postprocessing models for generative applications involving both speech and text, is introduced. MERS reduces the entropy rate of a model while remaining as close as possible to the starting model. This is shown to produce simplified models that concentrate on the most common and characteristic behaviours, and provides a continuum of simplifications between the original model and zero-entropy, completely predictable output. As the tails of fitted distributions may be inflated by noise or empirical variability that a model has failed to capture, MERS's ability to concentrate on high-probability output is also demonstrated to be useful for denoising models trained on disturbed data. / <p>QC 20131128</p> / ACORNS: Acquisition of Communication and Recognition Skills / LISTA – The Listening Talker

The estimation problem of stochastic nonlinear parametric models is recognized to be very challenging due to the intractability of the likelihood function. Recently, several methods have been developed to approximate the maximum likelihood estimator and the optimal mean-square error predictor using Monte Carlo methods. Albeit asymptotically optimal, these methods come with several computational challenges and fundamental limitations. The contributions of this thesis can be divided into two main parts. In the first part, approximate solutions to the maximum likelihood problem are explored. Both analytical and numerical approaches, based on the expectation-maximization algorithm and the quasi-Newton algorithm, are considered. While analytic approximations are difficult to analyze, asymptotic guarantees can be established for methods based on Monte Carlo approximations. Yet, Monte Carlo methods come with their own computational difficulties; sampling in high-dimensional spaces requires an efficient proposal distribution to reduce the number of required samples to a reasonable value. In the second part, relatively simple prediction error method estimators are proposed. They are based on non-stationary one-step ahead predictors which are linear in the observed outputs, but are nonlinear in the (assumed known) input. These predictors rely only on the first two moments of the model and the computation of the likelihood function is not required. Consequently, the resulting estimators are defined via analytically tractable objective functions in several relevant cases. It is shown that, under mild assumptions, the estimators are consistent and asymptotically normal. In cases where the first two moments are analytically intractable due to the complexity of the model, it is possible to resort to vanilla Monte Carlo approximations. Several numerical examples demonstrate a good performance of the suggested estimators in several cases that are usually considered challenging. / <p>QC 20171128</p>