Network based prediction models for coupled transportation-epidemiological systems

The modern multimodal transportation system provides an extensive network for human mobility and commodity exchange around the globe. As a consequence these interactions are often accompanied by disease and other biological infectious agents. This dissertation highlights the versatility of network models in quantifying the combined impact transportation systems, ecological systems and social networks have on the epidemiological process. A set of predictive models intended to compliment the current mathematical and simulation based modeling tools are introduced. The main contribution is the incorporation of dynamic infection data, which is becoming increasingly available, but is not accounted for in previous epidemiological models. Three main problems are identified. The objective of the first problem is to identify the path of infection (for a specific disease scenario) through a social contact network by invoking the use of network based optimization algorithms and individual infection reports. This problem parallels a novel and related problem in phylodynamics, which uses genetic sequencing data to reconstruct the most likely spatiotemporal path of infection. The second problem is a macroscopic application of the methodology introduced in the first problem. The new objective is to identify links in a transportation network responsible for spreading infection into new regions (spanning from a single source) using regional level infection data (e.g. when the disease arrived at a new location). The new network structure is defined by nodes which represent regions (cites, states, countries) and links representing travel routes. The third research problem is applicable to vector-borne diseases; those diseases which are transmitted to humans through the bite of an infected vector (i.e. mosquito), including dengue and malaria. The role of the vector in the infection process inherently alters the spreading process (compared to human contact diseases), which must be addressed in prediction models. The proposed objective is to quantify the risk posed by air travel in the global spread of these types of diseases.