Modeling the dynamics of epidemics helps to propose control strategies based on pharmaceuticaland non-pharmaceutical interventions (contact limitation, lockdown, vaccination,etc). Hand-designing such strategies is not trivial because of the number of possibleinterventions and the difficulty to predict long-term effects. This task can be cast as an optimization problem where state-of-the-art machine learning methods such as deep reinforcement learning might bring significant value. However, the specificity of each domain|epidemic modeling or solving optimization problems|requires strong collaborationsbetween researchers from different fields of expertise. This is why we introduce EpidemiOptim, a Python toolbox that facilitates collaborations between researchers inepidemiology and optimization. EpidemiOptim turns epidemiological models and cost functions into optimization problems via a standard interface commonly used by optimization practitioners (OpenAI Gym). Reinforcement learning algorithms based on QLearning with deep neural networks (DQN) and evolutionary algorithms (NSGA-II) are already implemented. We illustrate the use of EpidemiOptim to find optimal policies fordynamical on-o lockdown control under the optimization of the death toll and economic recess using a Susceptible-Exposed-Infectious-Removed (SEIR) model for COVID-19. Using EpidemiOptim and its interactive visualization platform in Jupyter notebooks, epidemiologists, optimization practitioners and others (e.g. economists) can easily compare epidemiological models, costs functions and optimization algorithms to address important choicesto be made by health decision-makers. Trained models can be explored by experts and non-experts via a web interface.
This article is part of the special track on AI and COVID-19.
L'objectif de cet article est de retracer les principaux enseignements issus des expériences récentes réalisées à partir du jeu du dictateur. Le jeu du dictateur est un jeu économique stylisé dans lequel le rapport entre les acteurs de la négociation est complètement asymétrique : l'un des acteurs (l'offreur) fait une proposition de partage d'une somme forfaitaire que son partenaire (le bénéficiaire) est obligé d'accepter. Contrairement à la prédiction théorique, notre revue des données expérimentales indique tout d'abord que le dictateur transfère une partie non négligeable de la somme à répartir (de 20% à 40% selon les études) et que cette proportion est d'autant plus élevée que la distance sociale qui le sépare du récipiendaire est réduite. L'identification ou la communication, qui réduisent la distance sociale, auraient en particulier pour effet de sensibiliser l'offreur à la norme sociale d'équité. Nous montrons ainsi que le bénéficiaire peut avoir intérêt, sous certaines conditions, à révéler son identité, à communiquer ses préférences ou même à tenter d'influencer l'offreur en jouant notamment sur son aversion à la culpabilité. Nous analysons ensuite les tentatives récentes de modélisation du comportement individuel permettant de tenir compte des faits stylisés caractéristiques du jeu du dictateur.
Summary: This article deals with the regulation of activities that entails risk by means of regulatory standard and liability. The original assumption is that the probability that parties might escape liability is variable among the population. The use of each instrument separately is first considered. Under liability, the optimal damages are calculated. From this, it is shown that regulation is superior to liability if the harm is not too variable within the population and if the probability of suit is sufficiently variable within the population, and conversely. The use of both instruments at the same time is then analyzed. The optimal combination of a safety standard and a liability schedule is derived. From this, it is proved that a joint use of regulation and liability is always optimal and that the instruments should be designed in a less stringent manner when used jointly.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.