CREDIBILITY INTERVAL DESIGN FOR COVID19 REPRODUCTION NUMBER FROM NONSMOOTH LANGEVIN-TYPE MONTE CARLO SAMPLING - Archive ouverte HAL Accéder directement au contenu
Proceedings/Recueil Des Communications Année : 2022

CREDIBILITY INTERVAL DESIGN FOR COVID19 REPRODUCTION NUMBER FROM NONSMOOTH LANGEVIN-TYPE MONTE CARLO SAMPLING

Résumé

Monitoring the Covid19 pandemic is critical to design sanitary policies. Recently, reliable estimates of the pandemic reproduction number were obtained from a nonsmooth convex optimization procedure designed to fit epidemiology requirements and to be robust to the low quality of the data (outliers, pseudo-seasonalities, \ldots). Applied to daily new infection counts made public by National Health Agencies and centralized by Johns Hopkins University, robust estimates of the reproduction number for 200+ countries are updated and published every day. To further improve estimation procedures and also, and mostly, increase their usability by epidemiologists, the present work exploits the Bayesian paradigm and derives a new Monte Carlo method to sample from a nonsmooth convex a posteriori distribution. This new sampler stems from an original combination of the Langevin Monte Carlo algorithm with Proximal operators. Its relevance and practical efficiency to produce meaningful credibility intervals for the Covid19 reproduction number are assessed from several indices quantifying the statistics of the Monte Carlo chains, and making use of real daily new infection counts.
Fichier principal
Vignette du fichier
CovidCI_Eusipco_R.pdf (656.45 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03371837 , version 1 (08-10-2021)
hal-03371837 , version 2 (20-02-2022)
hal-03371837 , version 3 (13-06-2022)

Identifiants

Citer

Hugo Artigas, Barbara Pascal, Gersende Fort, Patrice Abry, Nelly Pustelnik. CREDIBILITY INTERVAL DESIGN FOR COVID19 REPRODUCTION NUMBER FROM NONSMOOTH LANGEVIN-TYPE MONTE CARLO SAMPLING. pp.2196-2200, 2022, Proceedings of the 30th European Signal Processing Conference (EUSIPCO),, ⟨10.23919/EUSIPCO55093.2022.9909547⟩. ⟨hal-03371837v3⟩
478 Consultations
131 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More