IRD in Vietnam has set up a multidisciplinary team of researchers from UMMISCO (Huynh Quang Nghi, Nguyen Ngoc Doanh, Alexis Drogoul, Benoit Gaudou, Arthur Brugière, Kevin Chapuis), MIVEGEC (Marc Choisy, Damien Philippon) and DIADE (Pierre Larmande), assisted by colleagues from INRAE (Patrick Taillandier) to design realistic spatial computer models, in GAMA, from the data provided by the government (census, epidemiological data) or obtained from private actors (Facebook data, mobile telephone data) in order to inform as quickly as possible the public health decisions taken by the Vietnamese authorities, in particular those linked to the impact of containment strategies when cases are detected.
The goal is to allow quick comparisons between strategies (individual confinement, by neighbourhood, by commune, by zones, establishment of quarantines, etc.), to assess their influence on the dynamics of transmission, to propose optimisations in terms of temporality of implementation or combination of strategies (for example, closure of schools combined with confinement by district).
The models are built and validated using two case studies documented in recent weeks in Vietnam: the commune of Son Loi, province of Vinh Phuc (Red River Delta), the first to have been completely closed for almost 3 weeks, and another commune, in Ben Tre (Mekong Delta), which suffered the same fate recently due to the discovery of cases.
The two models start from an initial situation (discovery of "cases" within a synthetic population statistically correlated to the real population and populating an environment reproducing the real built environment) and explore by successive simulations the effects of different policies and the temporality of their implementation on the spread of the epidemic, using simple indicators (number of cases, number of deaths, spatial containment, etc.).
Generically constructed, these models are intended to be used with any set of input data (geographic, cadastral, epidemiological and demographic) in order to be able to assess, in advance, which policy to put in place to stem the epidemic at best, knowing that a lot depends on the density of the population, the topology of the places to circumscribe or the ability of the authorities to enforce the rules: thanks to the agent-based approach proposed in GAMA, all these elements can be taken into account, possibly interactively, by allowing decision-makers to participate - as in serious games - in the execution of simulations.
The models can be consulted and downloaded freely from http://comokit.org