Alonso Gabriel Ogueda (George Mason University) will give a virtualmath bio seminar presentation on “Physics-Informed Neural Networks Approaches to Predict Dynamics of Epidemiological Models Incorporating Human Behavior”.
Zoom: https://iastate.zoom.us/j/96940388489?pwd=cztb7TbtXfSbIFS2X9oJWV3xsfT70H.1
Physics-Informed Neural Networks Approaches to Predict Dynamics of Epidemiological Models Incorporating Human Behavior
Alonso Gabriel Ogueda (George Mason University)
In this talk, we introduce Physics-Informed Neural Networks (PINNs) as a flexible machine learning framework for integrating mechanistic models and data in applications for biological systems. After showcasing PINNs formulation and implementation for a classical compartmental SIR model as an example, we will demonstrate the performance of PINNs for two real world applications that incorporate human behavior: (i) an implicit COVID-19 transmission model, where PINNs recover time-varying parameters consistent with reported epidemic curves, and (ii) an SEIR-type epidemic model with vaccination and time-dependent level of immunity and using PINNs to understand the implications of under-reporting, vaccine efficiency and social behavior on the post-pandemic spread. Along with the implementation we also discuss pedagogical insights for teaching PINNs using a literate programming approach with Python, highlighting computational workflows that connect differential equations, optimization, and machine learning. Our findings suggest that PINNs is a powerful research framework for epidemic modeling and provides an accessible entry point for training students in modern computational and applied mathematics