Reinforcement learning based feedback control of tumor growth by limiting maximum chemo-drug dose using fuzzy logic

In this paper, a model-free reinforcement learning-based controller is designed to extract a treatment protocol because the design of a model-based controller is complex due to the highly nonlinear dynamics of cancer. The Q-learning algorithm is used to develop an optimal controller for cancer chemotherapy drug dosing. In the Q-learning algorithm, each entry of the Q-table is updated using data from states, action, and reward. The action is the chemo-drug dose. The proposed controller is implemented on a four states mathematical model including immune cells, tumor cells, healthy cells, and chemo-drug concentration in the bloodstream. Three different treatment strategies are proposed for three young, old, and pregnant patients considering his/her age. Chemotherapy is used in all cases.  In the older patient, immunotherapy is also used for modifying the dynamics of cancer by reinforcing his/her weak immune system. A Mamdani fuzzy inference system is designed to limit the maximum chemo-drug dose by regarding the age of the patients. Simulation results show the effectiveness of the proposed treatment strategy. It is also shown that immunotherapy is necessary for finite duration cancer treatment in patients with a weak immune system. The used strategy is a model-free method which is the main advantage of this method.