MultistateModels

Monte-Carlo expectation-maximization algorithm for fitting multi-state semi-Markov models to panel data in Julia. The algorithm can fit parametric and semi-parametric transition intensities with a proportional intensity parameterization for covariate effects.

Installation

You can install the development version of MultistateModels from GitHub with:

using Pkg
Pkg.add(url="https://github.com/fintzij/MultistateModels.jl.git")

Testing

Run the test suite:

using Pkg
Pkg.test("MultistateModels")

By default, quick unit tests run (~2 min). For the full suite including statistical validation tests:

MSM_TEST_LEVEL=full julia --project=. -e 'using Pkg; Pkg.test()'

Extended Test Suite

For comprehensive stress testing and statistical validation, see the separate test package: MultistateModelsTests.jl

To use:

# Clone into this directory
cd /path/to/MultistateModels.jl
git clone https://github.com/fintzij/MultistateModelsTests.jl MultistateModelsTests

# Activate and run
cd MultistateModelsTests
julia --project=. -e 'using Pkg; Pkg.instantiate()'
julia --project=. -e 'using MultistateModelsTests; MultistateModelsTests.run_longtests()'