Project Grant 2436801

This Federal Project Grant award, provided by the National Science Foundation's (NSF) Mathematical and Physical Sciences program (CFDA 47.049), supports the development of a first-principles informed, data-enabled predictive digital twin framework for human physiology. The $432,000 award to Arizona State University, a Hispanic-serving institution, will advance techniques for integrating real-world data into physics-based models to create personalized digital representations of human metabolic processes, with a focus on glucose regulation for individuals with type 1 diabetes.

The key products and services delivered through this research project include: (1) developing advanced neural network structures to recover underlying physics-based models from limited, noisy real-world data; (2) creating novel parameterizations of black-box dynamics using neural networks with built-in stability and robustness; (3) integrating heterogeneous, scarce, and noisy real-world data into virtual first-principles representations; and (4) developing a framework for learning unmodeled dynamics. The resulting digital twin capabilities are expected to enable more effective decision-making and evaluation of new treatments and therapies, while also supporting regulatory science and advancing foundational techniques for digital twin development in biomedical and healthcare domains.