Modeling of Cancer Tumour Growth and Optimum Control of Chemotherapy Drug Doses

Abstract

Cancer is a leading cause of death in Bangladesh as elsewhere in the world. Huge effort and measures have been taken to control cancer even though an alarming increase in new cases is predicted by doctors and health organizations. Successful treatment requires clear understanding of disease and its progression, early detection, optimum drug scheduling and expert oncologists/clinicians having experience and knowledge in cancer domain. Moreover, optimum control of chemotherapy drug dose scheduling is a very challenging task where many treatment constrains and objectives are to be meet which are often found inherently in conflict. This thesis presents an investigation into modeling of cancer tumour growth and optimum control of chemotherapy drug doses. A novel optimum chemotherapy drug scheduling is designed based on clinical practice and expert knowledge. Cancer tumour growth models are extensively investigated and one of the growth models is incorporated as an integral part of the chemotherapy drug dose scheduler to observe the response of a dose administered which, in turn, is used to calculate next schedule. In order to capture knowledge of a number of expert clinicians and oncologists, fuzzy systems are designed and used as the core components of the optimum control of chemotherapy drug doses. The proposed optimum chemotherapy drug scheduler will act as a decision support system for the clinicians/oncologist. Objective of this decision support system is to generate optimum chemotherapy drug dose scheduling considering the constraints, mainly keeping the toxicity under control and the system has to be clinically relevant. Many researchers gave their efforts to develop chemotherapy drug dose model, but most of these are often criticized due to lack of clinical relevance or hardly understandable to the clinicians/oncologists, as a result those are not useful to the oncologist. With this view in mind present decision support system has been developed in two phases. In the first phase a clinically relevant fuzzy expert system (FES) for tumour growth modelling and optimum chemotherapy drug dose scheduler is developed and in the second phase a fuzzy expert system with Physiologically Based Pharmacokinetic (PBPK) model based feedback controlled Decision Support System (DSS) for the oncologists is developed.

Clinically relevant fuzzy expert system for tumour growth modelling and optimum chemotherapy drug dose scheduler, a computational model, considers the patient’s Body Surface Area (BSA) and experts’ opinions to calculate chemo doses following the clinical practice. A proper balance between reducing cancerous cells and toxic side effects is required for effective drug scheduling. Still, in many cases, traditional clinical approaches fail to determine appropriate therapeutic doses that balance all restrictions. In the proposed system, Fuzzy Expert System-1 (FES-1) is developed to determine primary drug doses based on experts’ opinions and competing treatment objectives. To adjust the dose, Fuzzy Expert System-2 (FES-2) is developed based on clinical practices, the patient’s BSA, and experts’ opinions. The final chemotherapy drug dose schedule is generated by combining the outputs of FES-1 and FES-2, which is the proposed modular FES. A growth model is used in this work to observe response due to administration of chemotherapy drug doses and to determine the following doses by considering cancer patients’ three weight patterns (increasing, decreasing, and random order). Extensive simulation results and comparative assessment with other current computational chemotherapy drug scheduling models validate the effectiveness and the superiority of the model proposed in this study over the other methods reported in relevant studies. In the second phase, a fuzzy expert system with PBPK model based Feedback Controlled Decision Support System for the oncologists is developed. The system is developed to generate optimal chemotherapy drug doses with a view to reduce the cancerous tumour cells to zero, if possible, keeping the toxicity within the limit. Apart from these, the DDS model will also give the oncologists the opportunity to observe the drug concentration in different organs of the patient body. This will help/support the experts to justify the drug doses to be applied physically. The DSS model consists of three modules, which are a) Fuzzy Expert System for Chemotherapy Drug Dose Scheduler, b) Physiologically Based Pharmacokinetic (PBPK) Module for Drug Concentration Projection and c) Feedback Controller. The drug dose generated by the Fuzzy Expert System module will be inputted to the PBPK module to observe the drug concentration in different organs of the patient body, at the same time the feedback controller will help the fuzzy expert system to adjust the patient specific drug dose considering the output of the PBPK module and oncologist’s suggested/expected drug concentration in different organs of the patient body. The first module, clinically relevant fuzzy expert system, generates optimum drug doses, moreover in the present DSS model there is a provision to input threshold value of drug concentration for one or more organs considering the patient’s physical condition (kidney function, liver function, etc.) and the type of cancer, as a result more patient specific and cancer specific optimum chemotherapy drug dose projection is possible. All the doses generated by the proposed drug scheduling schemes could reduce the cancerous cells to almost zero while keeping the toxic side effects within the safe limit.