Bridging Scales in Cancer Modeling: Hybrid Framework, Surrogate Reduction, and Treatment Optimization

Despite advances in oncology, cancer persists as a major global health burden, with treatment efficacy often limited by incomplete understanding of its multiscale dynamics spanning tissues, cells, and molecular processes. To address this challenge, we developed a hybrid multiscale model that combines continuous descriptions at the tissue and molecular levels with a discrete individual-based approach at the cellular scale. To optimize chemotherapy treatment within this multiscale framework, we first derived a low-fidelity surrogate model using a data-driven approach integrating sparse identification of nonlinear dynamics and global sensitivity analysis. This reduced model enables efficient application of optimal control techniques for treatment optimization. Our framework provides a promising strategy for designing effective therapeutic protocols while minimizing drug toxicity in multiscale cancer scenarios.