We present a mathematical model that captures the transitions among three experimentally observed estrogen sensitivity phenotypes in breast cancer cells. Based on this model, a population-level model is created and used to explore the optimization of a therapeutic protocol