A Bayesian Phase I/II Trial Design


New Member
Dear all,
I have difficulties implementing the work written by Yuan Y, "A Bayesian Phase I/II Trial Design for Immunotherapy",

he proposed a Bayesian phase I/II dose-finding design that incorporates the unique features of immunotherapy by simultaneously considering three outcomes: immune response, toxicity, and efficacy. The objective is to identify the biologically optimal dose, defined as the dose with the highest desirability in the risk-benefit tradeoff. An Emax model is utilized to describe the marginal distribution of the immune response. Conditional on the immune response, we jointly model toxicity and efficacy using a latent variable approach.

I want after implementing this work to select the best dose in the first stage and to continue in a second stage with only 2 arms and the Best dose.

Yuan's work:

the code attached there is not complete, I run it for 50 iterations, it took 15 hours and I got two txt file (attached), but I do not know how he produces the results from these files,
he start with doses= c(.1,.3,.5,.7,.9)
Uti <- array(0,c(2,3,2)) # order: tox, eff, immuno
Uti[,,1] <- matrix(c(0,0,50,10,80,35),nrow=2)
Uti[,,2] <- matrix(c(5,0,70,20,100,45),nrow=2)
with oh.size=3 and N = 60

how he produces these figures and tables 2, Could anyone help, please?



Last edited:


Less is more. Stay pure. Stay poor.

Unfortunately, there aren't really any regulars on this site that would be familiar with this modeling. You may luck out and a random person answers it. I would have been your best bet, but I have only ran one dosing model before and it has been awhile. I will note, it is sad that that the modeling is taking so long, since I usually prefer to have say a thousand Bayesian sims so you can check the convergence and get credible intervals that may not be variable due to the modeling space.

Though, I am busy enough with work right now that I don't have time to read the article and try to figure this out. If you don't get much help here there is always Stacked Overflow/ Cross Validated, which have more traffic, but few of those posted questions get answered. You can also think about reaching out to the corresponding author on the paper.