Actualités

Journée Workshop « CoVeille »

Retour Accueil / Actualités / Journée Workshop « CoVeille »

Programme

- 10h Accueil café : Hall PG2
- 10h25-10h30 : Welcome address
- 10h30 – 11h30 : Vincent Brault (LJK, Université Grenoble Alpes) « Utilisation du pooling pour les tests RT-qPCR »
- 11h30 – 12h30 : Florence Débarre (IEES, Sorbonne Université) « How big is a cluster when you detect a first case ? and other Covid-19-inspired questions »
- 12h40 – 14h20 : Déjeuner Hall PG2
- 14h30 – 15h30 : Angelina Roche (CEREMADE, Université Paris Dauphine)
« Time-series forecasting with ABC prediction uncertainty and Reduced Order Models »
- 15h30 – 16h30 : Alejandro Mesejo-Chong (Universidad de La Habana)
« Forecasting the COVID-19 epidemic trend in Cuba with mechanistic and phenomenological models. Experiences from the University of Havana in the midst of the pandemic »

Abstracts

Vincent Brault « Utilisation du pooling pour les tests RT-qPCR »

L’une des problématiques de la pandémie actuelle de COVID-19 est la nécessité de pouvoir tester le plus largement possible les populations afin de mieux détecter la propagation et l’évolution. Toutefois, des problèmes techniques ont été mis en avant comme la tension sur la disponibilité des réactifs. Pour limiter ce problème, les méthodes de pooling (mélange de plusieurs échantillons avant de faire le test) sont régulièrement considérées en RT-qPCR (voir par exemple Gollier et Gossner (2020)).
Dans cet exposé, nous commencerons par expliquer en quoi consiste un test RT-qPCR et ce que cela implique sur les faux positifs et négatifs. Nous verrons ensuite le principe du pooling et comment cette procédure influence les résultats sur le taux de faux négatifs ; nous verrons en particulier l’importance de connaître la distribution de la concentration en charge virale. Nous continuerons donc sur la difficulté d’estimer cette concentration et nous conclurons par quelques procédures qui pourraient être appliquées pour aider en cette période de crise.


Florence Débarre « How big is a cluster when you detect a first case ? and other Covid-19- inspired questions »

Emerging epidemics and local infection clusters are initially prone to stochastic effects that can substantially impact the early epidemic trajectory. While numerous studies are devoted to the deterministic regime of an established epidemic, mathematical descriptions of the initial phase of epidemic growth are comparatively rarer. Here, we review existing mathematical results on the size of the epidemic over time, and derive new results to elucidate the early dynamics of an infection cluster started by a single infected individual. We show that the initial growth of epidemics that eventually take off is accelerated by stochasticity. As an application, we compute the distribution of the first detection time of an infected individual in an infection cluster depending on testing effort, and estimate that the SARS-CoV-2 variant of concern Alpha detected in September 2020 first appeared in the UK early August 2020. We also compute a minimal testing frequency to detect clusters before they exceed a given threshold size. These results improve our theoretical understanding of early epidemics and will be useful for the study and control of local infectious disease clusters.

Angelina Roche « Time-series forecasting with ABC prediction uncertainty and Reduced Order Models »

Sophisticated epidemiological models can include a large number of unknown parameters. This is true in particular of compartmental models such as extensions of the well-known Susceptible- Infectious-Recovered (SIR) model, which can be represented by a system of PDEs with up to 20 scalar parameters. These models are of interest as they are grounded in expert knowledge, but fitting them to data is intractable either for computational reasons or because detailed enough data are unavailable.
Bakhta et al. (2021) propose a model reduction technique, which projects the model with a large number of (constant in time) parameters onto a space with a small number of time-dependent parameters. A point prediction is then computed, based on a large number of simulated trajectories. We adapt ideas from the Approximate Bayesian Computation literature to leverage prediction uncertainty based on those simulated trajectories.
Joint work with Olga Mula (Univ. Paris-Dauphine), Robin Ryder (Univ. Paris-Dauphine), Ludovica Saccaro (Univ. Bordeaux), Giulia Sambataro (Univ. Bordeaux).

Alejandro Mesejo Chiong « Forecasting the COVID-19 epidemic trend in Cuba with mechanistic and phenomenological models. Experiences from the University of Havana in the midst of the pandemic »

As of March 2020, Mathematics as a science has gained an unexpected prominence in Cuba. The term "mathematical model’’ has become common in Cuban small and meaningful talks. We explain the why and how of this situation. Aspects of mathematical modeling of epidemics with deterministic methods will also be addressed.