HealthyMobility: optimal policies for human mobility to control COVID-19-epidemic spread
Date:
Changed on 09/07/2020
The HealthyMobility project aims to optimize the socio-economic constraints of a target containment, while respecting health constraints. Explanations.
What is the genesis of the HealthyMobility project?
The HealthyMobility project is the result of a collaboration between the Necs-Post team (CNRS, Gipsa, UGA, Inria) and the Mamba team (UMPC, Inria).
The Necs-Post team works in the field of modelling, estimation and control of large-scale systems, in the framework of Carlos Canudas de Wit's ERC Scale-FreeBack, which deals in particular with mobility and road traffic applications.
The Mamba team is interested in the modelling of medical and biological applications, and more specifically in epidemiology models.
From these complementary competences was born the idea of dealing with the problem of mobility control to optimize the socio-economic constraints of a target containment, while respecting health constraints.
How is it developing today and what are its objectives?
While deconfinement has been underway in France for more than a month, the HealthyMobility project aims to design optimal control strategies for human mobility for metapopulation models, or compartmentalized models with separate variables for the various categories of infection, with the aim of minimizing the infection rate by controlling mobility from areas of residence to social destinations (work, schools, shopping centers, restaurants, entertainment venues, administrations and health services).
The model considered is a multi-scale network model, where the mobility scale (hours) and the epidemic spreading scale (days) are interlinked. Control (or "gating") policies are aimed at minimizing the social and economic costs of mobility restrictions, but also at minimizing the possibility of resurgence of epidemic waves, while maintaining infection rates compatible with the capacity of intensive care units (ICUs).
By controlling mobility, i.e. the flow of people and its distribution during the day, the HealthyMobility project aims to minimize the spread of the epidemic, while meeting the economic cost and keeping peak infection rates below critical values.
Another innovative element of the project is the inclusion in the models of the number of tests (PCR type) applied during the day. This will also make it possible to imagine optimal testing strategies combined with mobility strategies.
Initially, the new models will be tested in simple scenarios to evaluate the impact of mobility and porting strategies to different destinations in the dynamics of the spread of the epidemic. In a second step, the model will be adapted to the Grenoble region, where the Necs-Post research team is based.
The project's progress
The team set out to define a model based on :
- multi-scale metapopulation epidemic models. Modified SIR-type models (here called SIDRA models) for the propagation of the epidemic under the effect of mobility from residential areas (communes) to other destination areas (work, leisure, shopping centres, etc.), including two different means of motorised transport (private and public),
- economic models and health constraints. Models representing the economic cost as a function of the rate of staff reduction will be used as a cost to be optimized, under constraints related to the epidemic and the occupancy of ICUs.
The model will then be calibrated using publicly available data.
Since some of the variables in the model cannot be measured, observers (virtual sensors) will be designed to reconstruct the unmeasured variables.
Finally, we will study optimal gating (specific deconfinement rate for each destination) and optimal testing policies.
Eventually, the algorithms will be integrated into an open web platform that can be used as a guide for public decision making during the reopening process of different cultural institutions and venues. The platform will allow online simulations for various model parameters and a variety of economic and societal objectives.