Resumen
In this paper, we consider an application of robust model predictive control to optimal dosing of breast cancer chemotherapy. The model-patient mismatch is handled by computing an ellipsoidal invariant set containing the measured patient's states at each sampling time. An optimal dose of chemotherapeutic agent is obtained by solving a convex optimization problem subject to linear matrix inequalities. In the case study of simulated patients, the results show that the tumor volume can be reduced to a specified target with up to 30% model-patient mismatch. Moreover, the robust model predictive control algorithm can achieve better treatment results as compared with the nonlinear model predictive control algorithm while the on-line computational time is significantly reduced.