A data-driven model inversion approach to cancer immunotherapy control

A novel data-driven control design approach for Multiple Input Multiple Output nonlinear systems is proposed in the paper, relying on the identification of a polynomial prediction model of the system to control and its on-line inversion. A simulated study is then presented, concerning the design of a control strategy for cancer immunotherapy. This study shows that the proposed approach may be quite effective in treating cancer patients, and may give results similar to (or perhaps better than) those provided by “standard” methods. The fundamental difference is that “standard” methods are typically based on the unrealistic assumption that an accurate physiological model of the cancer-immune mechanism is available; in the approach proposed here, the controller is designed without such a strong assumption.