trialr (version 0.0.7)

model_EffTox: Stan model for EffTox dose-finding design

Description

This is the Stan implementation of the EffTox design for dose-finding in scenarios where dose decisions are guided by efficacy and toxicity outcomes.

The model is compiled when trialr is installed and is available at run-time under stanmodels$EffTox.

The design was first published by Thall & Cook in 2004. Advice on its use was materially updated in 2014.

See the EffTox vignette for a detailed description of the probability model and a demonstration of the pertinent methods implemented in trial.

References

Thall & Cook. 2004, Dose-Finding Based on Efficacy-Toxicity Trade-Offs

Thall, Herrick, Nguyen, Venier & Norris. 2014, Effective sample size for computing prior hyperparameters in Bayesian phase I-II dose-finding

See Also

efftox_params

efftox_parameters_demo

efftox_process

Examples

# NOT RUN {
# Get example parameters from Thall et al. 2014
dat <- efftox_parameters_demo()
# Add outcomes for three patients
dat$num_patients <- 3
dat$eff <- c(0, 1, 1)
dat$tox <- c(0, 0, 1)
dat$doses <- c(1, 2, 3)
# Fit the observed data to the model using rstan
fit <- rstan::sampling(stanmodels$EffTox, data = dat)
# The fit object is quite crude. Post-process to perform useful inference:
decision <- efftox_process(dat, fit)
decision$recommended_dose == 3
# Dose-level 3 is recommended for the next cohort of patients.
# }