Action Space

The control-space in this problem is the insulin infusion rate provided by the selected Insulin Pump. Hence, the action-space of the problem is a continuous one dimensional action space.

The insulin infusion rates vary for the different insulin pump. These information will be updated here in the documentation shortly.

Please refer to our paper on “Non-linear Continuous Action Spaces for Reinforcement Learning in Type 1 Diabetes”, where we analyse different action space formulations suitable for this problem.