Dengue research Framework for resisting Epidemics in Europe

Vector-borne diseases are a major public health burden with a high potential for epidemic outbreaks, particularly in regard to climate change and the increasing urbanization worldwide. The mosquito Aedes aegypti is one of the main vectors of such diseases that include Dengue, Zika and Chikungunya. Pending effective vaccines, the presence of this mosquito species, which is perfectly adapted to urban environments, needs to be monitored and controlled. This includes the production of geographic databases and dynamic models contributing to the assessment of vector and epidemic risks at different scales, especially the micro-localization of urban spaces. These risks arise from an articulation of environmental, biological and social factors that we proposed to considere as a “complex pathogenic system”. In this context we have developed a vector-borne disease simulator to explore two important issues: the reproductive and dispersal capacity of Aedes aegypti according to local socio-environmental contexts and the role of human mobility in the dispersion of viruses transmitted by the mosquito at intra-urban scales. Three sub-models have been developped and explored separately in order to calibrate them: MOMA, MODE and MOMOS, respectively model of mosquito, of environment and host.

MOMA (Model Of Mosquito Aedes) is a spatially explicit agent-based simulation model of Aedes aegypti female mosquito. The model aims to produce statistical data on mosquito behaviours and population dynamics that are difficult to obtain through field surveys such as population densities in various geographical and climatic conditions. It can also be used to explore effects of vector control strategies on population dynamics. The model simulates adult mosquitoes as ‘agents’ which interact with their local environment. The latter provides resources for their biological development and can also constrain their flight or egg-laying behaviours. Variations in environmental configurations such as land-use and climate make it possible to explore the dependence of mosquito population dynamics on the context. Different approaches have been used to calibrate and validate it.

Studying dynamical and non-linear relationships between the environment and vectors provides us new hypotheses. So we implemented a more sophisticated model of environment in order to generalise some of the main environmental concepts previously developped and to study more systematically those interactions.

MODE’s (Model of Dynamical Environement) objective is to identify environmental dynamics that contribute to the spatiotemporal distribution of Aedes aegypti in urban areas. We used the resource-based habitat concept to develop a generic method for estimating the environmental factors of this habitat and their dynamics. These concepts and methods were employed using different data from Bangkok (Thailand) and from Delhi (India). We were able to estimate local urban population and land-use using remote sensing image. We also developped a methodology to estimate the daily minimum and maximum air temperatures in an urban setting. Finally we developped methods to estimate the potential breeding site inside and outside households and their dynamics in terms of water capacity.

MOMOS’ (Model Of MObility Simulation) aims to simulate urban daily mobility at the scale of individual. Doing so, we are able to reproduce micro scale interactions between humans and vectors to explore the possible dynamics of virus progression in changing environmental contexts. We have therefore developed an agent-based model to generate a synthetic population. Its first objective is to be socially and statistically representative in terms of the composition of the population. We added data to describe the individual mobilities and to calibrate the behavior of the model so that it is convergent with the real mobilities observed in Bangkok. We have used data from a social network (Twitter) which is an interesting alternative to the more traditional mobility data. These data enabled us to produce individual agendas of mobility and global statistic on mobility in Bangkok.

A fourth element called DBVirus concerns data base used as an input for the contagion model between host and vector and for the infection dynamics. This database concern information regarding dengue virus and its caracteristic both for host and for mosquito: extrinsec-intrinsec incubation, viremia, probability of symptomatic-asymptomatic case etc.

These four elements (MOMA, MOMOS, MODE and DBVirus) was studied separately to estimate their capacities to produce known parameters and to produce original information:

MOMA: Study of simulated mosquito behaviours reveals the model’s ability to produce the mosquito’s realistic life cycle and life expectancy. The mosquito cohort’s flight distance in various urban landscapes was also explored. Our simulation results reveal a significant relation between urban topology, built topography, human densities and the mosquito flight dispersal. We used this important result by producing risk maps of mosquito dispersal in different urban contexts.

MOMOS: We were able to identify strong correlations between Twitter data and inhabitants at their places of residence. This good spatial representativity of our sample was then used to produce statistical information on daily flow of population in Bangkok. We were able to identify major places of attraction in the urban area and the average distance covered by individuals to carry out an activity (school, shopping, work etc.) according to their place of residence. We also have produced individual agendas for typical days in the week and in the weekend. Simulations of hundreds of thousands agents following these agendas reproduce the daily mobility of Bangkok.

MODE: Estimation of air temperature using surface temperatures in Bangkok makes it possible to get satisfying results on the whole. These air temperature estimation methods appear to be profitable in a context of vector control in inter-tropical zone countries. They are likely to allow identification of risk areas wherein the thermal conditions remain highly favorable for maintaining Aedes aegypti populations as well as the viruses that they transmit, during the coldest periods. We use these estimates, and precipitation, as input in the model. An evaporation model has been integrated in order to capture the level of water into the breeding sites, which acts on the egg-pupae-larva development stages.

After calibration and validation, these 3 submodels have been integrated in a single model called MO3. This model enables us to simulate life cycle of millions of mosquitoes. In the simulator, according to meteorological data and the presence of breeding sites, mosquito can lay their eggs; In the presence of humans, they can take blood meal. At the same time, according to their agenda, human move on from place to place, every two hours. During interactions between host and vector, if one of them is infected by dengue virus, then transmission can occur. As a result, it is now possible to follow step by step the path of the virus diffusion. MO3 allows us to follow the population dynamic of the vectors at the city scale in order to map the main hot-spots and their environmental features. MO3 allows us to map the main hot-spots of contamination thanks to the record of each event during the simulation. MO3 is then now suitable to explore many hypothesis on dengue virus diffusion at city scale, depending on factors such as seroprevalence in the population, densities of mosquito population, targeted source reduction.