| Abstract: |
| Dengue fever is a vector-borne viral disease primarily transmitted by Aedes mosquitoes. Both population structure and dengue transmission are highly influenced by climate conditions. In this talk, I will present a dynamical system model and a math-model-informed neural network (MINN) based method to predict climate-driven spatiotemporal dynamics of mosquito populations. Furthermore, I will develop a hybrid probabilistic-mechanistic, data-driven model that enables us to estimate a practical, local-level risk of dengue infection. We use data from Nepal, which provides a valuable setting for dengue modeling due to its strong spatial heterogeneity; its 77 districts span a wide range of elevation and climate zones. We analyze our model to formulate reproduction numbers that determine the global dynamics of mosquito survival. Our method identifies the climate conditions that help control dengue-transmitting mosquitoes. Our results provide critical insights into the role of climate change in shifting the distribution of dengue-transmitting mosquitoes and related epidemics into colder regions. Our model`s real-time risk assessment provides an evidence-based methodology for designing public health policies. |
|