Master thesis presentation in Physics. The thesis title is "Modelling cancer dynamics during therapy from liquid biopsy measurements".
Liquid biopsy measurements make it possible to monitor the progression of solid cancers in almost real-time. By simulating the interaction between tumour growth and treatment, mathematical models can be used to predict the tumour evolution. In this thesis, multiple mathematical models were fitted to liquid biopsy measurements of neuroblastoma patients. The models ranged from well-known and relatively simple models, such as Logistic or Bertalanffy, to more complex ones, which take two sub-populations and resistance or regrowth into account. Different inference methods, including maximum likelihood estimation and hierarchical modelling, were compared. Additionally, we checked if the models were structurally identifiable using synthetic data.
We showed that multiple-population models fit the data significantly better compared to single-population ones. This highlights the complexity of the disease and could explain for the bad prognosis related to it. In particular, the Fast-Slow Exponential model offers a simple model that outperforms all tested 1D-models. This model assumes that there are two sub-populations, which both decay exponentially under treatment, but at different rates.
Supervisor: Eszter Lakatos
Examiner: Sviatlana Shashkova and Marija Cvijovic