Dual-Agent Dose Escalation for Phase I Trials using the PIPE Design

Implements the Product of Independent beta Probabilities dose Escalation (PIPE) design for dual-agent Phase I trials as described in Mander AP, Sweeting MJ (2015) .


  • New web browser interface using Shiny; runShinyPIPE()
  • Corrected a bug where the dimensions of the dose-space were incorrectly calculated if parameters a and b were given as argument. Many thanks to Roxane Duroux for spotting this mistake.
  • pipe.design object now return a and b, the parameters of the beta prior for each dose-combination
  • plot.pipe allows an extra plot to be produced that gives the empirical probabilities of DLT (posterior medians) and 95% credible intervals
  • Densities of the posterior probability of toxicity at each dose level can be calculated using the distribution of monotonic contours. These can be plotted after each cohort is recruited.
  • Fixed a bug where program crashes if prior.med is not specified (thanks to Jing Hu for reporting this)
  • pipe.design returns an object "rpII.list" listing all the recommended Phase II dose combinations for each simulation
  • Allows non-diagonal escalation in combination with a neighbouring or no.dose.skip constraint
  • Uses a new optimsation algorithm to calculate beta a and b parameters from median and sample size. Previous version did not converge for large sample sizes.
  • Clarify in the help files that the recommendation percentages are the percentage of times each dose combination is recommended over the total number of recommended combinations in the simulated trials (which is often more than N)
  • First release.

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.5.1 by Michael Sweeting, 5 months ago

Browse source code at https://github.com/cran/pipe.design

Authors: Michael Sweeting

Documentation:   PDF Manual  

Task views: Design of Experiments (DoE) & Analysis of Experimental Data

GPL (>= 2) license

Imports ggplot2, gtools, xtable

Suggests shiny

See at CRAN