Examining the Effect of Initial Parameter Value Choice of a Tumor Model on Gradient-Based Parameter Fitting

Cyber-medical systems are becoming more and more popular in health sciences. They help to make faster and more accurate diagnoses and create more effective and personalized treatments for patients. During our research, we use a mathematical model that describes the behavior of the tumor and the effect of the drug for therapy personalization. Our goal is to personalize the parameters in the mathematical model and optimize therapy based on this. As a result of the optimization, we aim to treat cancer with smaller doses of chemotherapy administered at more frequent intervals, resulting in milder side effects and a lower chance of developing drug resistance. In this article, we examine three parameters of the tumor model to find the parameter range of the initial values such that the identification algorithm finds the real value. We created virtual patients whose parameters were obtained by fitting a model from previous experiments; thus, the parameters are known, and the results of the identification can be evaluated.