We study fast-slow versions of the SIR, SIRS and SIRWS epidemiological models, and of the SIRS epidemiological model on homogeneous graphs, obtained through the application of the moment closure method. The multiple time scale behavior is introduced to account for large differences between some of the rates of the epidemiological pathways. Our main purpose is to show that the fast-slow models, even though in nonstandard form, can be studied by means of Geometric Singular Perturbation Theory (GSPT). In particular, without using Lyapunov's method, we are able to not only analyze the stability of the endemic equilibria of the SIR and SIRS models, but also to show that in the remaining models limit cycles arise. We show that the proposed approach is particularly useful in more complicated (higher dimensional) models such as the SIRWS model and the SIRS on homogeneous graphs, for which we provide a detailed description of their dynamics by combining analytic and numerical techniques. In particular, for the latter we show that the model can give rise to periodic solutions, differently from the corresponding model based on homogeneous mixing.

Tout ce qui vit, naît, se nourrit, se reproduit et meurt. Pour le cerveau, la question se complexifie car à la survie des neurones s'ajoute le coût de l'activité cérébrale. La question de la gestion énergétique pour les neurones est particulière car les cellules de notre cerveau évoluent de manière concertée et non par compétition. On sait avec l'imagerie médicale que l'usine neuronale ne fonctionne pas uniquement grâce au glucose ; elle utilise d'autres apports énergétiques tels que le lactate ou le glutamate pour soutenir sa production. Lorsqu'une tumeur apparaît, elle change le métabolisme énergétique pour survivre et soutenir sa propre croissance. En particulier, les cellules cancéreuses se fournissent en lactate et choisissent leur substrat préféré en fonction de l'oxygène disponible. La modélisation mathématique des substrats énergétiques est un outil de choix pour décrire et prédire de tels flux. Coupler ces modèles à des données issues de l'IRM et de la SRM permet d'améliorer la prise en charge du patient présentant un gliome.Cette thèse propose l'approche de plusieurs dynamiques en substrat dans le cerveau sain et gliomateux en se basant sur des systèmes d'équations : échanges locaux en lactate (EDO, système lent-rapide), échanges globaux en substrats (EDO), cycle glutamate/glutamine (EDR) et échanges en lactate en dimensions supérieures (EDP). Ces modèles sont expliqués, décrits grâce aux mathématiques et permettent l'élaboration de simulations ajustées selon des données patient ou issues de la littérature.L'énergie est nécessaire au maintien de la vie. Mais si votre voisin consomme une partie de vos ressources, pouvez-vous encore espérer survivre ? / Everything that lives is born, eats, reproduces and dies. For the brain, the question is more complex because neurons have to survive and to support brain activity. Energy management is also particular because brain cells evolve together with no competition. Thanks to medical imaging, we know that neurons do not consume only glucose. They can use others energetic substrates such as lactate and glutamate as a power source.When a tumor appears, it changes the energetic metabolism to survive and support its own growth. In particular, cancer cells like to consume lactate. They also choose their favorite substrate based on the available oxygen. Modeling of energy substrates is useful to describe and predict energetic kinetics and changes. Mathematical models could get with clinical and medical results to describe, explain or predict low grade glioma dynamics. They can help to characterize and quantify a tumor evolution, then leading to improve their therapeutical management. Exchanges between mathematics and MRI (and MRS) enable to get accurate data and to build suitable mathematical models.This thesis deals with several approaches of substrates dynamics in healthy and gliomatous brains. These researches are based on systems of equations. We model local lactate exchanges (ODE, fast-slow systems), global substrates exchanges (ODE), glutamate/glutamine cycle (RDE) and local lactate exchanges in higher dimensions (PDE). We describe, analyze and give simulations of these models. Simulations are fitted on patient MRI data or literature data. Energy is necessary to live. But if your neighbor consumes a part of your resources, can you still survive ? / Tutto ciò che vive nasce, si nutre, si riproduce e muore. Per il cervello, la questione è più complessa perché i neuroni devono sopravvivere e sostenere l'attività cerebrale. La gestione energetica cerebrale è particolare anche perché le cellule cerebrali evolvono insieme, senza concorrenza. Inoltre, grazie alle immagini mediche, sappiamo che i neuroni non consumano solo del glucosio ma usano altri substrati energetici come il lattato o il glutammato.Quando un tumore si stabilisce, cambia il metabolismo energetico del cervello per sopravvivere e sostenere la propria crescita. In particolare, cellule tumorali consumano del lattato e scelgono il loro substrato preferito basandosi all'ossigeno disponibile.La matematica, e in particolare l'elaborazione di modelli matematici può aiutarci a ottimizzare i dati disponibili, che possono essere, di volta in volta, delle proprietà cellulare o delle lastre MRI o MRS. La modellizzazione dei substrati energetici potrebbe descrivere, spiegare o prevedere le dinamiche energetiche nel cervello.Questa tesi tratta di diversi approcci della dinamica dei substrati nei cervelli sani e gliomatosi. Queste ricerche si basano su sistemi di equazioni. Modellizziamo scambi locali di lattato (ODE, sistemi fast-slow), scambi globali di substrati (ODE), ciclo glutammato/glutammina (RDE) e scambi locali di lattato in dimensioni superiori (PDE). Descriviamo, analizziamo e diamo simulazioni di questi modelli. Le simulazioni sono adeguate su dati MRI paziente o dati di letteratura.Per vivere, l’energia è una necessità. Ma se i Suoi vicini consumassero le Sue risorse, riuscirebbe ancora a sopravvivere ?