Hi
JointSurvivalModels.jl is a package for Bayesian joint modeling of survival and longitudinal data. You can define distributions based on a hazard function with links to longitudinal models (and covariates).
It subtypes the ContinuousUnivariateDistribution type from Distributions.jl and implements the calculation of the probability density function / likelihood of observations and the generation of a random sample of your joint survival distribution. Besides the type JointSurvivalModel , this package also defines the abstract type HazardBasedDistribution which implements the numeric calculations. Given a user-defined hazard function it allows you to calculate various distribution functions and generate random samples.
The motivating application for this software was Bayesian inference frameworks, such as Turing.jl, to achieve Bayesian modeling of joint models with non-linear mixed effects models for the longitudinal observations. This requires a numeric approach to the likelihood calculation and for the generation of random samples. It was initially developed for oncology research and now made open source.
I welcome any feedback, ideas, or contributions!