When applying machine learning methods to real world problems, there is often the need for configuring a variety of model-related variables and algorithm hyperparameters. However, this is not always a straightforward task, and interactive approaches must be incorporated, where the end-user can perform small-scale experiments and directly monitor the interim results of the algorithms.
This can be achieved with graphical user interfaces (GUI) that interoperate with the algorithms' backend and that allow access to internal states and variables, which should also be easily re-configurable by the end-user. This thesis is based on an application scenario from the life sciences domain, in particular on model exploration for tumor treatments.