Speakers
Description
Cancer is a complex multiscale disease characterized by heterogeneous cellular behaviors, nonlinear dynamics, and interactions across scales. Computational models provide a formal framework to study cancer progression, predict treatment response, and support translational research. However, significant challenges remain due to spatial and temporal scales, nonlinear behavior, and the high-dimensional, heterogeneous data that are difficult to integrate into mechanistic frameworks. Machine learning approaches easily integrate multi-modal data across diverse spatial and temporal scales to make classifications and predictions but often this comes at the cost of interpretability that is standard in mathematical oncology. This minisymposium aims to bring together researchers to discuss recent advances in integrating machine learning with mechanistic models for cancer research in (1) hybrid approaches combining mechanistic models and machine learning; (2) machine learning methods for calibration and adaptation of mechanistic models; (3) mechanistic learning approaches to improve interpretability, robustness, and generalization; and (4) machine learning models for knowledge discovery, pattern extraction, and guidance of new mechanistic models. Applications span in vitro, in vivo, and patient-specific digital twins for personalized cancer progression prediction, and treatment planning. [1] Rockne, et al. "The future of mathematical oncology in the age of AI." npj Sys Bio & Appl (2026).
