Data-driven Cancer Modeling And Optimization Of Treatment Strategie

Breast cancer and prostate cancer are two common types of diagnosed malignancy in the female and male populations,respectively,and the incidence of these two diseases has been high in China in recent years.A effective treatment to prevent or delay the progression of cancer has been the kernel of researches by scholars.This paper focuses on the evolution of tumor and the optimized treatment scheme by mathematical models and optimization algorithms for breast cancer and prostate cancer.Excessive accumulation of β-catenin proteins is a key driver in the development of breast cancer.In this paper,we develop novel mathematical models for breast cancer,available clinical data are used to guide construction of models and consequent selection of therapeutic regimens in a faster manner and lower cost.Firstly,we establish a breast cancer model on mice without treatment to explore the interaction between cancer cells and immune system within the tumor microenvironment.Secondly,the model without treatment is modified in order to incorporate the effects of RNAi treatment and immunotherapy,respectively.Furthermore,the gradient descent method and particle swarm algorithm are designed to optimize therapy schemes in order to inhibit the growth of tumor and lower the treatment cost.Considering the mechanisms of drug resistance,simulations exhibit that therapies are ineffective resulting in cancer relapse in the prolonged time.For this reason,parameter sensitivity analysis shows that injection of mature dendritic cells has potential for advancement as a new anti-tumor strategy.Intermittent androgen suppression therapy,which inhibits the growth of androgenindependent tumor cells,has been widely used to treat prostate cancer patients.We first define the union set of disjoint closed intervals in the real number field as a new time scale,then a prostate cancer model is established on the time scale,sufficient conditions for exponential stability and global uniform asymptotic stability of the model are obtained by using stability theory.Secondly,the model parameters are determined by particle swarm optimization coupled with simulated annealing,and the influence of parameters for cancer relapse is analyzed theoretically.Finally,given the clinical data of two patients,the original regimens are optimized by minimizing the largest eigenvalue,and the results show that the optimized treatment regimens shorten the treatment process,reduce the treatment cost and delay the relapse.