<p>Actual cancer treatment is typically a dynamic process with stochastic disturbances. Uncertain constraints (UCs) are suitable for modeling of dynamic processes under stochastic disturbance conditions, in which constraints are not fully met. Therefore, uncertain constrained dynamic optimization (UCDO) models can be utilized for addressing anti-cancer drug administration (ACDA) in cancer treatment. Due to the dynamics, randomness, and complexity of decision functions, the UCDO problem arising from ACDA in cancer treatment is difficult to deal with. To tackle this issue, a relaxation technique (RT) and an improved smooth approximation strategy (ISAS) are proposed for formulating the UCDO problem as a deterministic approximation problem, where a vector parameterization strategy and equality/inequality constraint dealing with method are integrated. Following that, to attain a global optimal solution (GOS) to the deterministic approximation problem, a hybrid optimization method (HOM) is proposed based on limited memory BFGS (L-BFGS) and a novel stochastic search method (NSSM) and its global convergence results are established. Simulation results show that the proposed HOM can achieve a higher quality solution with a lower calculating cost and lower conservativeness than existing approaches for solving the ACDA problem in cancer treatment under stochastic disturbances.</p>