Bayesian Individual-based Compartmental Inference (BICI)
Compartmental models have long been used as a means of understanding the collective dynamics of interacting agents, with notable applications in epidemiology and ecology. BICI allows for arbitrary compartmental model specification (using an easy point and click interface) and performs simulation and/or inference on that model.
The current version (v1.0) supports the following features:
Arbitrary compartmental models. The interface allows for easy model specification and can accommodate multiple classifications (e.g. disease status as well as location and sex of individuals).
Markovian and Non-Markovian transitions. This allows for more realistic models (e.g. disease recovery can be modeled using a more adaptable gamma distribution instead of assuming an exponential distribution).
Simulation. The initial conditions can be set and the dynamic variation in the model can be graphically represented in a variety of ways.
Inference. BICI can take a variety of different data types (e.g. state data, population estimates, event times or even uncertain data such as disease diagnostic test results) and infer model parameters as well as underlying model dynamics.
Variation in time. The software allows for the the possibility of model parameters discretely varying in time. Also variation in parameters with the age of individuals can be incorporated.
BICI is provided freely for anyone to use under a GNU General Public License. We only request that those who use BICI analysis in their publications cite this tool.
BICI runs as a desktop application and can be download by clicking on one of the following links (the downloaded file needs to be extracted and the software is run by clicking on the BICI icon):
(Windows)
(Mac)
(Linux)
