Predicting mosquito response to visual targets with Bayesian dynamical systems inference

2023 saw nine cases of non-travel-related mosquito-borne malaria in the United States. In sub-Saharan Africa, malaria kills 600,000 people every year, most of them children under 5; approximately one child every minute. Despite years of research in into mosquito host seeking, a quantitative understanding of their behavior remains elusive. Here, we perform 3D infrared tracking of the Aedes aegypti mosquitoes in an environmental chamber at the Center for Disease Control and use Bayesian dynamical systems inference methods to learn quantitative models for their behavior in response to sensory cues. We focus on their host selection criteria based on visual cues by providing mosquitoes with a pair of different-sized black spheres to simulate different-sized hosts. We find that mosquitoes are more attracted to larger spheres and darker colors according to our mathematical model of their visual system. Quantitative models of mosquito behavior learned from 3D tracking experiments may provide important insight into mosquito host selection and inspire the design of more effective mosquito traps.