In urban and peri-urban environments throughout tropical and subtropical regions of the world, immature stages of Aedes aegypti mosquitoes (eggs, larvae and pupae) develop in water-holding containers. A number of detail-rich models have been developed to couple the dynamics of the immature and adult stages of Ae. aegypti, with temperature playing an important role in mortality, development and egg-laying rates, and density-dependent mortality affecting larvae. The numerous assumptions of these models enable them to realistically characterize impacts of mosquito control on Aedes-borne diseases (e.g. dengue, chikungunya, Zika), but they also constrain such models’ ability to reproduce empirical patterns that do not conform to the models’ behavior. In stark contrast, modern statistical methods afford sufficient flexibility to extract nuanced signals from noisy data, yet they have very limited ability to make predictions about impacts of mosquito control on disease without extensive data on both. Here, we demonstrate how the differing strengths of mechanistic realism and statistical flexibility can be fused into a single model. Our analysis utilizes data from 176,352 household-level Ae. aegypti aspirator collections conducted during 1999-2011 in Iquitos, Peru.
Results/Conclusions
The key step in our approach is to calibrate a single parameter of the model to spatio-temporal abundance patterns predicted by a generalized additive model (GAM). This calibrated parameter essentially absorbs residual variation in the abundance time-series not captured by the mechanistic model, fusing its dynamics with those predicted by the statistical model. We performed this calibration in the deterministic analog of the final agent-based model, using literature-derived estimates, alongside temperature data from Iquitos, for the remaining mortality and development rates. The parameter we calibrated was the rate of pupal and larval mortality induced by routine container maintenance, although theoretically any parameter could be chosen for this purpose.
We then used this calibrated parameter and the literature-derived parameters in the agent-based model to explore Ae. aegypti population dynamics and the impact of insecticide spraying to kill adults. The baseline abundance predicted by the agent-based model closely matches that predicted by the GAM. Following spraying, agent-based model predicted abundances rebound within 30 days, commensurate with recent experimental data from Iquitos. Our approach is able to accurately reproduce abundance patterns in Iquitos and produce a realistic response to adulticide spraying, while retaining sufficient flexibility to be applied across a range of settings.