Reinforcement control with fuzzy-rules emulated network for robust-optimal drug-dosing of cancer dynamics

In this article, a nonlinear mathematical model of the biological phenomena in chemotherapy cancer treatment is considered as a class of unknown discrete-time systems when the input data and the measured output are only available. The input data are the drug administration represented as the control effort and the output is the tumor cells population. As a result, the actor-critic architecture is constructed without the full-state observer. Two sets of IF-THEN rules are utilized for fuzzy rules emulated networks by human knowledge according to the pharmacokinetic and pharmacodynamic details. The learning laws are derived from the concept of the incoherent reward function. Thus, the convergence of the internal signals and the robustness are accomplished by the theoretical and numerical results. Furthermore, the comparative results are given to demonstrate the effectiveness of the proposed scheme.