Summary

The National Science Foundation's Division of Mathematical Sciences awarded $465,777 to the University of California, Irvine under the Mathematical and Physical Sciences program (CFDA 47.049) to develop hypothesis-based interpretability methods for high-dimensional machine learning models of T cell signaling. The project, active from July 1, 2026 through June 30, 2029, integrates classical mathematical insights from decades of immunology research with modern artificial intelligence and machine learning techniques to achieve enhanced predictive accuracy for immune cell responses. The research specifically addresses T cell responses to complex, multi-frequency signals and co-signaling receptor combinations—areas previously challenging to model due to the high dimensionality of simultaneous signal interactions.

Beyond research outputs, the grant deliverables include graduate student training in machine learning, immunology, and applied mathematics, as well as development of a new course on coding practices for reliable use of modern artificial intelligence coding assistants. The project aims to advance quantitative, predictive modeling of cellular signal processing with particular relevance to cell-based immunotherapies, while addressing critical challenges in model interpretability, computational efficiency, and overfitting in high-dimensional biological systems.