Optimal control and computational subtyping: a dual framework for understanding and treating cancer

This dissertation integrates control theory modeling and spatial phenotyping to address the dual challenges of treatment resistance and diagnostic precision in oncology. By simulating tumor evolutionary dynamics, the work identifies treatment schedules that manage subpopulation diversity to delay resistance more effectively than traditional reduction-based strategies. Furthermore, a machine learning framework leverages spatial proteomic data to map unique tumor-immune architectures, achieving 95% accuracy in breast cancer subtype classification. Together, these frameworks provide computational tools to optimize adaptive therapy and refine spatial diagnostics in precision medicine.