JointSurvivalModels.jl

This package implements joint models in Julia. It was designed to support the modeling of joint models with probabilistic programming, for example using the Turing.jl framework. Install it with:

using Pkg
Pkg.add("JointSurvivalModels")

The JointSurvivalModel implements a canonical formulation of joint models. It based on a hazard function $h(t) = h_0(t) \exp(\gamma' \cdot L(M(t)))$ with a baseline hazard $h_0$ and a link to joint models $L(M(t))$. For a more detailed explanation see the documentation.

The hazard of the exponential distribution $\text{Exp}(\alpha)$ is the constant function $x\mapsto \alpha$. For the joint longitudinal model we use a simple cosinus function. The joint hazard is then $h(t) = \alpha \exp(\gamma * \cos(t))$.

using JointSurvivalModels
constant_alpha(x) = 0.2
γ = 0.5
jm = JointSurvivalModel(constant_alpha, γ, cos)

Plotting the survival function vs the baseline hazard:

using StatsPlots, Distributions
r = range(0,12,100)
plot(r, ccdf(Exponential(1/0.2), r), label="Baseline survival")
plot!(r, ccdf(jm, r), label="Joint Survival")

For a more instructive example take a look at the documentation first example or the case study found in example/.

Contributions are welcome, the issue tracker is a good place to start.

This project is licensed under the terms of the MIT license.