bayesTSM

Bayesian 3-state AFT model for screening outcomes.

Please note:

• This 'package' currently is a collection of functions. We plan to publish an R library soon.
• The main function is bts_survreg
• Use Rstudio project or define working directory path
• To load all functions required source path /controllers/startup.r
• For a demonstration of the Bayesian model estimation by Gibbs check the demo_MCMC.r with demonstrations of bts_survreg using simulated data
• For a demonstration of the EM algorithm for the lognormal-lognormal model, see demo_EM.r
• The foreach package is used for parallel and sequential processing. If your system is incompatible with foreach consider using the bts_survreg_seq function instead which uses a for loop insead (can be slow).

Author information: Thomas Klausch (t.klausch@amsterdamumc.nl). Reference as: Klausch, T., Akwiwu, E. U., van de Wiel, M. A., Coupé, V. M. H., Berkhof, J. (2021). A Bayesian accelerated failure time model for interval censored three-state screening outcomes. Available under: https://arxiv.org/abs/2110.02649

License: This product is provided under a creative commons license: Attribution-NonCommercial-ShareAlike CC BY-NC-SA.

Important note: Throughout the code and output, the annotation X is used for the time from state 1 to state 2 and S is used for the time from state 2 to state 3. In the accompanying main paper the notation $x$ and $t$ are used for these times, respectively. Letters t and T are problematic to use in R as they have frequently used other definitions. To avoid conflicts the notation S is used throughout.