Ce travail de thèse s'articule autour de trois parties distinctes abordant chacune un thème précis lié à l'épidémiologie. La première partie de ces travaux s'inscrit dans le cadre de la propagation de virus via l'utilisation de modèles épidémiques. Dans cette partie, sont analysées différentes méthodes d'estimations paramétriques et y sont étudiés la qualité de ces estimateurs. Une application à des virus informatiques est proposée. La deuxième partie de cette thèse propose une méthode d'estimation de la prévalence actuelle du virus de l'hépatite C en France par l'intermédiaire d'un modèle de rétro-calcul associé à un modèle de Markov modélisant l'histoire naturelle de la maladie. Cette méthode et les résultats qui en découlent sont comparés avec les résultats obtenus via l'approche de référence en France. Enfin, la dernière partie s'intéresse à l'étude de l'impact des nouvelles thérapeutiques anti-hépatite C susceptible d'éradiquer le virus à moyen terme. En assimilant la population d'intérêt à un groupement de graphes aléatoires, la propagation du virus est modélisée à partir d'un modèle de métapopulation construit sur la base de données migratoires où les dynamiques de chaque sous-population sont régies par un ensemble d'équations différentielles déterministes. Ce travail doctoral a été réalisé dans le cadre d'une convention CIFRE avec les laboratoires Bristol-Myers Squibb. / The works undertaken in this doctoral thesis are conducted in three parts, each one dealing with a specific epidemiology-related domain. The first part of this work deals with the propagation of viruses by using well-known epidemic models. It is mainly focused on the analyze of different estimation methods and on their performance. An application on computer virus is proposed. The second part of this thesis gives an estimation method of the hepatitis C virus prevalence in France based on a back-calculation model in association with a Markov model of the disease's natural history. This method and its results are compared with those generated by the reference approach in France. The last part is focused on the study of the recent anti-hepatitis C therapeutics impact on the population since is has been stated that those could eradicate the virus at middle term. In that optic, based on published migration data and assuming that the population of interest is organized into a set of specific contact networks, a metapopulation is computed in which the dynamics of each sub-population is governed by a set of deterministic differential equations. This doctoral research has been conducted through a CIFRE industrial research agreement with the Bristol-Myers Squibb pharmaceutical company.

With the algorithm’s suitability for exploiting current petascale and next-generation exascale supercomputers, stable and structure-preserving properties are necessary to develop predictive computational tools. In this dissertation, summation-by-parts (SBP) operators and a new relaxation Runge–Kutta (RRK) scheme are used to construct mimetic and structure-preserving full discretization for non-reactive compressible computational fluid dynamics (CFD) and reaction-diffusion models. In the first chapter, we provide the necessary background and a literature survey that forms the basis of this dissertation. Next, we provide a short overview of entropy stability for general conservation laws. The second chapter covers the analysis of the Eulerian model for compressible and heat-conducting flows. We provide the necessary background of the new system of parabolic partial differential equation (PDE). Then, we present the entropy stability analysis of the model at the continuous level. Subsequently, using the SBP, we construct an entropy-stable discretization of any order for unstructured grids with tensor-product elements. The third chapter discusses the implementation of RRK methods. We start by reviewing the RRK scheme constructed to guarantee conservation or stability with respect to any inner-product norm. Then, we present the extension and generalization of RRK schemes to general convex functionals and their application to compressible fluid flow problems. The final chapter demonstrates the far-reaching capabilities of the SBP operators and RRK schemes presenting the development of a novel fully discrete Lyapunov stable discretization for reaction models with spatial diffusion. Finally, we conclude this dissertation with an overview of our achievements and future research directions.

Modelos matemáticos de motoneurônios têm sido desenvolvidos para auxiliar na compreensão dos fenômenos que envolvem o sistema neuromuscular. Entretanto, a maioria dos modelos já desenvolvidos baseou-se na premissa de que a árvore dendrítica tem um comportamento passivo, o que ocorre em animais anestesiados, mas pode não ocorrer durante o comportamento motor normal de um animal intacto. Experimentos com animais descerebrados, em que as vias monoaminérgicas encontravam-se ativas, mostraram que os motoneurônios podem apresentar comportamentos mais complexos decorrentes da presença de condutâncias iônicas voltagem-dependentes que se situam nos dendritos e são responsáveis pela gênese de uma corrente de entrada persistente. Nesse sentido, um primeiro objetivo deste trabalho foi o de desenvolver novos modelos matemáticos de motoneurônios de diferentes tipos (i.e. dos tipos S, FR e FF), computacionalmente eficientes e contendo em seus compartimentos dendríticos uma condutância de cálcio do tipo L, de forma que os fenômenos de biestabilidade, potencial platô e amplificação da corrente sináptica efetiva possam ser gerados. Um segundo objetivo foi o de verificar como a presença da condutância iônica ativa no dendrito influencia o comportamento motoneuronal quando o mesmo está sujeito a entradas sinápticas de diferentes tipos. Os novos modelos foram parametrizados baseando-se em dados da literatura experimental para motoneurônios de gatos descerebrados e validados segundo os protocolos experimentais básicos que permitem caracterizar cada tipo de modelo como sendo totalmente ou parcialmente biestável. As entradas sinápticas foram simuladas por processos pontuais de Poisson e os trens de potenciais de ação dos motoneurônios foram analisados. Uma modulação senoidal da intensidade do processo pontual foi usada para estimar as respostas em frequência de cada modelo. Observou-se que, funcionalmente, a presença da condutância iônica dendrítica pode favorecer a ação do motoneurônio durante tarefas posturais, pois, uma vez ativada, a corrente de entrada persistente eleva a excitabilidade motoneuronal tornando os disparos mais regulares, além de prover uma alta sensibilidade dos modelos a entradas sinápticas de baixa frequência, correspondentes às oscilações observadas durante a manutenção da postura ereta quieta. / Mathematical models of motoneurons have been developed as an aid to the understanding of phenomena involving the neuromuscular system, but most of these models have been based on the hypothesis of a passive dendritic tree. This holds for anesthetized animals but not necessarily during normal motor behavior of the intact animal. Experiments with decerebrate animals in which the monoaminergic tracts were maintained intact have shown that more complex behaviors may emerge in motoneurons due to dendritic voltage-gated ionic conductances, which are responsible for a persistent inward current. Therefore, the first aim of this work was to develop computationally-efficient new motoneuron models of different types (i.e. type S, FR and FF) that include a dendritic L-type calcium conductance so that bistability, plateau potential and enhancement of effective synaptic current may be generated. The second aim of this research was to evaluate the effects of the active dendritic ionic conductance on the input-output mapping of presynaptic to postsynaptic spike trains. The new models were parameterized based on data reported in experimental literature on the decerebrate cat preparation, and they were validated using appropriate protocols for either fully or partially bistable dynamics. The synaptic inputs were simulated by Poisson point processes and the output spike trains were analyzed. Sinusoidal modulation of the point process intensity was used for the estimation of each models frequency response. The results suggested that an active dendritic ionic conductance in motoneurons has a functional role during postural tasks, because, when activated, the persistent inward current enhances the motoneuronal excitability, reducing the variability of interspike intervals, and focusing the sensitivity of the models to low frequency inputs that correspond to the low-frequency oscillations that typically occur during quiet standing posture.

Modelos matemáticos de motoneurônios têm sido desenvolvidos para auxiliar na compreensão dos fenômenos que envolvem o sistema neuromuscular. Entretanto, a maioria dos modelos já desenvolvidos baseou-se na premissa de que a árvore dendrítica tem um comportamento passivo, o que ocorre em animais anestesiados, mas pode não ocorrer durante o comportamento motor normal de um animal intacto. Experimentos com animais descerebrados, em que as vias monoaminérgicas encontravam-se ativas, mostraram que os motoneurônios podem apresentar comportamentos mais complexos decorrentes da presença de condutâncias iônicas voltagem-dependentes que se situam nos dendritos e são responsáveis pela gênese de uma corrente de entrada persistente. Nesse sentido, um primeiro objetivo deste trabalho foi o de desenvolver novos modelos matemáticos de motoneurônios de diferentes tipos (i.e. dos tipos S, FR e FF), computacionalmente eficientes e contendo em seus compartimentos dendríticos uma condutância de cálcio do tipo L, de forma que os fenômenos de biestabilidade, potencial platô e amplificação da corrente sináptica efetiva possam ser gerados. Um segundo objetivo foi o de verificar como a presença da condutância iônica ativa no dendrito influencia o comportamento motoneuronal quando o mesmo está sujeito a entradas sinápticas de diferentes tipos. Os novos modelos foram parametrizados baseando-se em dados da literatura experimental para motoneurônios de gatos descerebrados e validados segundo os protocolos experimentais básicos que permitem caracterizar cada tipo de modelo como sendo totalmente ou parcialmente biestável. As entradas sinápticas foram simuladas por processos pontuais de Poisson e os trens de potenciais de ação dos motoneurônios foram analisados. Uma modulação senoidal da intensidade do processo pontual foi usada para estimar as respostas em frequência de cada modelo. Observou-se que, funcionalmente, a presença da condutância iônica dendrítica pode favorecer a ação do motoneurônio durante tarefas posturais, pois, uma vez ativada, a corrente de entrada persistente eleva a excitabilidade motoneuronal tornando os disparos mais regulares, além de prover uma alta sensibilidade dos modelos a entradas sinápticas de baixa frequência, correspondentes às oscilações observadas durante a manutenção da postura ereta quieta. / Mathematical models of motoneurons have been developed as an aid to the understanding of phenomena involving the neuromuscular system, but most of these models have been based on the hypothesis of a passive dendritic tree. This holds for anesthetized animals but not necessarily during normal motor behavior of the intact animal. Experiments with decerebrate animals in which the monoaminergic tracts were maintained intact have shown that more complex behaviors may emerge in motoneurons due to dendritic voltage-gated ionic conductances, which are responsible for a persistent inward current. Therefore, the first aim of this work was to develop computationally-efficient new motoneuron models of different types (i.e. type S, FR and FF) that include a dendritic L-type calcium conductance so that bistability, plateau potential and enhancement of effective synaptic current may be generated. The second aim of this research was to evaluate the effects of the active dendritic ionic conductance on the input-output mapping of presynaptic to postsynaptic spike trains. The new models were parameterized based on data reported in experimental literature on the decerebrate cat preparation, and they were validated using appropriate protocols for either fully or partially bistable dynamics. The synaptic inputs were simulated by Poisson point processes and the output spike trains were analyzed. Sinusoidal modulation of the point process intensity was used for the estimation of each models frequency response. The results suggested that an active dendritic ionic conductance in motoneurons has a functional role during postural tasks, because, when activated, the persistent inward current enhances the motoneuronal excitability, reducing the variability of interspike intervals, and focusing the sensitivity of the models to low frequency inputs that correspond to the low-frequency oscillations that typically occur during quiet standing posture.

L’être humain est organisé selon une horloge interne d’une période d’environ 24 heures. La pharmacocinétique de certaines classes de médicaments est donc influencée par le rythme circadien. En effet, l’aire sous la courbe de la concentration en médicament en fonction du temps, la concentration maximale en médicament et le temps auquel on obtient la concentration maximale peuvent varier en fonction de l’heure à laquelle a été consommé le médicament. Le but de ce travail est de modéliser la variation de la concentration maximale de ces médicaments selon le moment de la journée auquel ils sont pris. On étudie d’abord un modèle présenté par Godfrey permettant de trouver la concen- tration en médicament en fonction du temps et tenant compte des variations circadiennes. Ce modèle ne permet pas d’illustrer les variations dans la concentration maximale selon le moment de la journée auquel le médicament est pris. Un nouveau modèle à deux com- partiments sera donc développé pour les trois modes d’absorption (orale, intraveineuse, intraveineuse bolus). Les systèmes d’équations différentielles résultants seront étudiés. L’effet de la variation des paramètres de phase sur la concentration maximale sera aussi étudié. La preuve de l’existence des solutions, de leur unicité et de leur positivité sera faite en annexe. / Humans are organised according to an internal clock with a period of approximatively 24 hours. The pharmacokinetic of several classes of drugs are then influenced by circadian rhythms. Indeed, the area under the curve (of the drug concentration as a function of time), the maximal concentration and the time to maximal concentration can change according to the time at which the drug is taken. The objective of this present work is to find a model to represent the variations in the maximal drug concentration according to the absorption’s time. We first study a model presented by Godfrey. It allows to find the drug concentration as a function of time while taking into account circadian rhythms. Unfortunately, this model could not represent the variations in the maximal concentration according to the time at which the drug is taken. We developed a new two-compartmental model for the three ways of absorption (oral, intravenous and intravenous bolus). The resulting systems of ordinary differential equations will be studied. The effect of the phase parameters on the maximal concen- tration will also be studied. Finally, the proof of well-poseness of the model will be developed in the Annex.

La dengue est une infection virale qui touche de 100 à 400 millions d'individus chaque année. Selon l'OMS, « la dengue sévère est l’une des principales maladies graves et causes de décès dans certains pays d’Asie et d’Amérique latine ». Il est justifiable de modéliser la propagation de cette maladie dans une population à l'aide de modèles mathématiques compartimentaux. Les travaux de Forshey et al. sur la fièvre dengue semblent indiquer la possibilité qu'une infection à la dengue ne donne pas une immunité à long terme contre les différents sérotypes du virus, et qu'une réinfection homotypique à la dengue serait commune. Nous étudions un modèle SIRS de la dengue qui prend en compte cette perte d'immunité via un système d'équations différentielles à délai. Nous caractérisons les états stationnaires et leur stabilité en termes des différents paramètres considérés, notamment les taux de reproduction de base associés à chacun des sérotypes de la dengue. Nous étudions les bifurcations du système en ses principaux paramètres, notamment les bifurcations de Hopf émergeant de la présence d'un délai dans le système d'équations différentielles. Des simulations numériques du modèle sont présentées afin de représenter les différents régimes du modèle à l'étude. / Dengue is a viral infection affecting from 100 to 400 million people each year. According to the WHO, "severe dengue is a leading cause of serious illness and death in some Asian and Latin American countries". This justifies the modelling of this illness's propagation in a population using mathematical compartmental models. Results of Forshey et al. on dengue fever seem to indicate the possibility that a dengue infection does not yield a long term immunity against the different dengue serotypes, and that an homotypical reinfection could be common. We study a SIRS model for the dengue virus that takes into account this loss of immunity via a system of delay differential equations. We characterize the stationary states and their stability in terms of the different parameters considered, in particular the basic reproduction ratios associated to each dengue serotype. We study the system's bifurcations in its main parameters, especially the Hopf bifurcations arising from the presence of a delay in the system of differential equations. Numerical simulations of the model are presented to represent the model's different regimes.

Dans la première partie de cette thèse, nous avons mis en place un métamodèle de transmission du paludisme basé sur la modélisation compartimentale susceptible-infecté-résistant (SIR) et prenant en compte les flux de mobilité humaine entre différents villages du Centre Sénégal. Les stratégies d’intervention géographiquement ciblées, s’étaient avérées efficaces pour réduire l’incidence du paludisme aussi bien dans les zones d’intervention qu’à l’extérieur de ces zones. Cependant, des actions combinées ciblant à la fois le vecteur et l’hôte, coordonnées à large échelle sont nécessaires dans les régions et pays visant l’élimination du paludisme à court/moyen terme.Dans la deuxième partie nous avons évalué différentes méthodes d’estimation de la mobilité humaine en l’absence de données individuelles. Ces méthodes incluaient la traçabilité spatio-temporelle des téléphones mobiles ainsi que les modèles mathématiques de gravité et de radiation. Le transport de l’agent pathogène dans l’espace géographique, par la mobilité d’un sujet infecté est un déterminant majeur de la vitesse de propagation d’une épidémie. Nous avons introduit le modèle d’impédance qui minimise l’erreur quadratique moyen sur les estimations de mobilité, en particulier dans les contextes où les ensembles de population sont caractérisés par leurs tailles hétérogènes.Nous avons enfin élargi le cadre des hypothèses sous-jacentes à la calibration des modèles de gravité de la mobilité humaine. L’hypothèse d’une distribution avec excès de zéros a fourni un meilleur ajustement et une meilleure prédictibilité, comparée aux hypothèses classiques n’assumant pas un excès de zéros : Poisson, Quasipoisson. / In the first part of this thesis, we have developed a malaria transmission metamodel based on the susceptible-infected-resistant compartmental modeling framework (SIR) and taking into consideration human mobility flows between different villages in the Center of Senegal. Geographically targeted intervention strategies had been shown to be effective in reducing the incidence of malaria both within and outside of intervention areas. However, combined interventions targeting both vector and host, coordinated on a large scale are needed in regions and countries aiming to achieve malaria elimination in the short/medium term.In the second part we have evaluated different methods of estimating human mobility in the absence of real data. These methods included spatio-temporal traceability of mobile phones, mathematical models of gravity and radiation. The transport of the pathogen through the geographical space via the mobility of an infected subject is a major determinant of the spread of an epidemic. We introduced the impedance model that minimized the mean square error on mobility estimates, especially in contexts where population sets are characterized by their heterogeneous sizes.Finally, we have expanded the framework of assumptions underlying the calibration of the gravity models of human mobility. The hypothesis of a zero inflated distribution provided a better fit and a better predictability, compared to the classical approach not assuming an excess of zeros: Poisson, Quasipoisson.

The dynamics of many aspects of social behaviour, such as spread of fads and fashion, collective action, group decision-making, homophily and disagreement, have been captured by mathematical models. The power of these models is that they can provide novel insight into the emergent dynamics of groups, e.g. 'epidemics' of memes, tipping points for collective action, wisdom of crowds and leadership by small numbers of individuals, segregation and polarisation. A current weakness in the scientific models is their sheer number. 'New' models are continually 'discovered' by physicists, engineers and mathematicians. The models are analysed mathematically, but very seldom provide predictions that can be tested empirically. In this work, we provide a framework of simple models, based on Lotka's original idea of using chemical reactions to describe social interactions. We show how to formulate models for social epidemics, social recovery, cycles, collective action, group decision-making, segregation and polarisation, which we argue encompass the majority of social dynamics models. We present an open-access tool, written in Python, for specifying social interactions, studying them in terms of mass action, and creating spatial simulations of model dynamics. We argue that the models in this article provide a baseline of empirically testable predictions arising from social dynamics, and that before creating new and more complicated versions of the same idea, researchers should explain how their model differs substantially from our baseline models. / Matematiska modeller kan hjälpa oss att förstå många typer av sociala fenomen, som ryktesspridning, spridning av memes, gruppbeslut, segregation och radikalisering. Det finns idag otaliga modeller för sociala beteenden hos människor och djur, och fler presenteras kontinuerligt. Det stora antalet modeller försvårar navigering inom forskningsfältet, och många av modellerna är dessutom komplicerade och svåra att verifiera genom experiment. I detta arbete föreslås ett ramverk av grundläggande modeller, som var och en modellerar en aspekt av socialt beteende; det gäller sociala epidemier, cykler, gemensamt handlande, gruppbeslut, segregation och polarisering. Vi menar att dessa modeller utgör majoriteten av de verifierbara aspekter av socialt beteende som studeras, och att de bör behandlas som en utgångspunkt när en ny modell ska introduceras. Vilka av mekanismerna från utgångspunkten finns representerade i modellen? Skiljer den sig ens nämnvärt från utgångspunkten? Genom att ha en god förståelse för grundmodellerna, och genom att förklara på vilket sätt en ny modell skiljer sig från dem, kan forskare undvika att presentera modeller som i praktiken är mer komplicerade varianter av sådana som redan finns. I detta arbete visar vi hur dessa grundläggande modeller kan formuleras och studeras. Modellerna bygger på enkla regler om vad som händer när individer i en befolkning möter varandra. Till exempel, om en person som har vetskap om ett rykte träffar någon som inte har det, kan ryktet spridas vidare. Därför har antaganden om vilka personer som kan träffa varandra stor påverkan på de resultat som modellerna ger. I detta arbete studeras varje modell med två olika metoder: i den ena har alla personer i befolkningen samma sannolikhet att träffa varandra, i den andra representeras befolkningen av ett rutnät, där varje plats motsvarar en individ. I den senare har alltså varje person ett begränsat antal grannar att interagera med. Vilken av dessa två metoder man väljer har stor betydelse för vilka beteenden modellerna förutspår. Som ett komplement till detta arbete presenteras ett verktyg i form av ett Python-program som utför analysen av modellerna. Detta kan användas för att undersöka grundmodellerna som presenteras i detta arbete, men också för att formulera och analysera nya modeller på samma sätt. På det viset kan nya modeller enkelt jämföras mot grundmodellerna. Verktyget är användbart både som introduktion för de som är nya inom social dynamik, men också för de forskare som som vill ta fram nya modeller och föra forskningsfältet vidare.