Sandia National Laboratory Collaboration (Spring 2025, Fall 2025)
The MS in Applied Data Science program has partnered with Sandia National Laboratory for two semesters on a project exploring a hybrid modeling approach for epidemiological analysis. While traditional epidemiological models are easier to implement and enables the computation of key epidemiological parameters like the basic reproduction number and the potential duration of infectiousness, they ignore individual interactions and thus do not always represent data associated with the spread of the disease. Agent-based models are useful for studying individual interactions that influence the spread of the disease; however, they are computationally intensive. In this project, students explore data-driven modeling which incorporates machine learning and statistical methods within a differential equation modeling framework through the use of two mechanisms: universal differential equations and a neural ODE approach.
Students worked across two semesters on this project in teams. The resulting final presentations from each semester can be found below. Furthermore, one student extended this project into a Master’s thesis by using a stochastic differential equation in conjunction with a neural network. It can be found at the following link:
Alice Menaya Armah-Bonney, “Stochastic Universal Differential Equations for Epidemiological Modeling: Uncertainty Quantification in Disease Transmission Dynamics”: https://dc.etsu.edu/etd/4655/