Model-free feedback design for a mixed cancer therapy

In this article, a model-free feedback control design is proposed for the drug administration in mixed cancer therapy. This strategy is very attractive because of the important issue of parameter uncertainties unavoidable when dealing with biological models. The proposed feedback scheme use past measurements to update an on-line simplified model. The control design is then based on model predictive control in which a suitable switching is performed between two different cost functions. The effectiveness of the proposed model-free control strategy is validated using a recently developed model (unknown to the controller) governing the cancer growth on a cells population level under combined immune and chemotherapy and using real human data.