Description du projet
Une nouvelle théorie des processus épidémiques non markoviens
La crise du coronavirus a révélé un certain nombre de lacunes dans le domaine de l’épidémiologie traditionnelle. Ces études négligent, par exemple, le graphe des contacts humains et supposent implicitement une population homogène sans structure de graphe spécifique. Du point de vue de la physique, les deux processus les plus importants sont la façon dont le virus passe de A à B et la façon dont A et B entrent en contact. Alors que la plupart des études sur les épidémies en réseaux supposent implicitement des dynamiques de propagation de type markovienne, pour lesquelles les théories mathématiques sont bien développées, ce n’est pas le cas de la COVID-19. Dans ce contexte, le projet ViSioN, financé par le CER, développera une théorie non markovienne des processus épidémiques susceptible de déterminer combien de temps durera une pandémie et quand un pic se produira.
Objectif
ViSioN presents my Network Science view on virus spread in networks, in which the duality between the virus transmission process and the contact graph is key.
The devastating Corona crisis reveals two major shortcomings in traditional epidemiology. First, it ignores the human contact graph and implicitly assumes a homogeneous population without specific graph structure. Second, most models for the virus spreading process relate to a Markovian setting, with exponential infection and curing times, leading to an exponential decay of the epidemic. Measurements, however, point to significantly different infection and curing time distributions. In addition, digital technology can help in constructing the contact graph and combined with medical testing, all infected can be detected, thus avoiding a second wave.
Building on my pioneering work on Markovian epidemics in networks, I will complement the recipe book of epidemic model ingredients with corresponding algorithms/software for next pandemic outbreaks. I will develop the theory of non-Markovian epidemic process on networks, a surprisingly missing element today, because non-Markovian theory is needed to tell, based on the characteristic infection and curing times of the virus, how long a pandemic will last and when the peak occurs. Next, I will combine all available measurement technologies to construct the best possible contact graph via temporal networking or adaptive networking. Finally, I will explore how accurately infections can be predicted under partial information of process and contact graph.
ViSioN’s outcomes will allow to predict, manage and control any epidemic in the best possible way. Moreover, as epidemics are part of the larger class of “local rule–global emergence” systems, my outcomes will be directly beneficial for the other members of this broad and abundant class, and find applications ranging from computer malware spread to human brain surgery.
Champ scientifique
Programme(s)
Thème(s)
Régime de financement
ERC-ADG - Advanced GrantInstitution d’accueil
2628 CN Delft
Pays-Bas