C++ Implementation of the 2016 Chase ICING model
This project requires Boost 1.62.0 at the time of this writing. See CMakeLists.txt for the most up to date Boost requirements. You can obtain Boost from your package manager in Debian-based distributions using:
sudo apt install libboost-all-dev
CMake should take care of the rest!
This is a CMake project. To build the test executable, unfortunately you'll need access to a private repository.
If you did have access to the repository, the build would be as simple as:
mkdir build && cd build && cmake .. && make (on linux)
The most important aspect of this repo is not the test executable (which shows how to attach a controller to the model), but the model itself.
The model is a header-only library, contained in include/ChaseIcing.hpp. This contains a templated ChaseIcing class.
It can be instantiated in a default format using:
ChaseIcing<>::state_type initial_state = {...};
auto model = ChaseIcing<>(initial_state);
The model can be run for a specified number of minutes ([time_start, time_end]), with a specified update interval dt, and exogenous IV insulin flow insulin_rate (in mU/min) and exogenous IV dextrose flow dextrose_rate (in mmol/min) using ChaseIcing<>().run():
model.run(time_start, time_end, dt, insulin_rate, dextrose_rate)
The resulting glucose, in mmol/L, can be obtained using ChaseIcing<>().glucose():
auto resulting_glucose = model.glucose()
An example library, with a (mostly correct) initial state vector, can be found in lib/icing.cpp
In the future I plan to extend the model using randomly generated insulin sensitivity as derived from Fig 4 and Fig 5 of [1], to simulate patients properly from the original cohort of ICU patients. This will be implemented using the existing InverseCDFProcess<U> template class that's in the include folder.
[1] - TBME-01160-2016
[2] - computer methods and programs in biomedicine 102 (2011) 192–205