joint work with Jana de Wiljes, Niklas Hartung, Charlotte Kloft, Wilhelm Huisinga
For cytotoxic anticancer drugs, the combination of a narrow therapeutic window and a high variability between patients complicates dosing policies, as strict adherence to standard protocols may expose some patients to life-threatening toxicity and others to ineffective treatment. Therefore, individualisation is required for safe and effective treatments. We present a reinforcement learning framework for individualised dosing policies that not only takes into account prior knowledge from clinical studies, but also enables continued learning during ongoing treatment.
joint work with Manuel Tetschke, Sebastian Sager
The disease Polycythemia Vera (PV) leads to patients, which suffer from an uncontrolled production of red blood cells (RBC). If untreated, this might be fatal. Initial treatment is done by phlebotomy in certain regular intervals. Until now it is not known, how to find an individual optimal timing of the treatment schedule, w.r.t. side effects. We extend our published model of RBC dynamics to PV and verify it using simulations. Different approaches for the computation of optimal treatment schedules are formulated and discussed. We illustrate our conclusions with numerical results
joint work with Enrico Schalk, Daniela Weber, Hartmut Döhner, Thomas Fischer, Sebastian Sager
In this talk we present a pharmacokinetic-pharmacodynamic model describing the dynamics of healthy white blood cells (WBCs) and leukemic cells during chemotherapy for acute myeloid leukemia patients. Despite the primary treatment goal of killing cancer cells, a further clinically relevant event is the cytarabine-derived and lenograstim-reduced low number of WBCs increasing the risk of infections and delayed or stopped therapy. We present optimal treatment schedules derived by optimization problems considering terms for disease progression, healthiness and therapy costs.