The trialr package
trialr is an R package that implements Bayesian clinical trial design in Stan, a modern enviroment for probabilistic programming that implements Hamiltonian Monte Carlo Markov Chain sampling.
Bayesian models are relatively abundant in early phase clinical trials where information is scarce and sensible inference is required practically from the moment the trial begins.
- the Contimual Reassessment Method (CMR) for toxicity-based dose-finding;
- the time-to-event CRM for late-onset toxicity in dose-finding;
- the EffTox for toxicity- and efficacy-based dose-finding;
- Wason & Seaman's Augmented Binary method for fusing continuous and binary outcomes in phase II;
- a hierarchical model of binary responses on related subgroups;
- and more.
trialr is available on CRAN and is listed on my GitHub page. There is a manuscript currently under peer-review (pre-print linked below) and it has been used to analyse outcomes in a forthcoming set of clinical trial results.
- A Phase II Clinical Trial Design for Associated Co-primary Efficacy and Toxicity Outcomes with Baseline Covariates
- trialr - Bayesian Clinical Trial Designs in R and Stan
- How to design a dose-finding study using the continual reassessment method
- Combination Lenalidomide and Azacitidine - A Novel Salvage Therapy in Patients Who Relapse After Allogeneic Stem-Cell Transplantation for Acute Myeloid Leukemia
- Implementing the EffTox dose-finding design in the Matchpoint trial