This repository contains a diabetic patient data simulator following the structure of an open-ai gym and is originally intended to evaluate reinforcement learning algorithms.
Install this package via
pip install -e .
or
python setup.py install
Can be found in setup.py.
This simulator was inspired by the following work:
- Maintain Glucose in Type-I Diabetic
- simglucose, an implementation of the FDA-approved 2008 version UVA/Padova Simulator
This simulator is based on an expanded version of the Bergman minimal model, which includes meal disturbances. The underlying mathematical representation of this model was first developed by John D. Hedengren.
The goal is to keep glucose levels at a tolerable level in Type-1 diabetic patients. This process can be controlled using remote insulin uptake.
For additional details on this gym, see dosing_rl_gym/resources/Diabetic Background.ipynb.