Skip to contents

Summary the toxicity info based on MCMC samples from crmPack::mcmc()

Summary_crmPack_Samples.Rd

Summary the toxicity info based on MCMC samples from crmPack::mcmc()

Usage

Summary_crmPack_Samples(dose, ref_dose, samples, prob_cut)

Arguments

dose

a scalar of dose

ref_dose

reference dose in the logistic toxicity model

samples

MCMC samples from crmPack::mcmc()

prob_cut

a length-2 numeric vector of DLT rate cut. For the DLT rate:

  • [0, prob_cut[1]) is underdose interval;

  • [prob_cut[1], prob_cut[2]) is target dose interval;

  • [prob_cut[2], 1] is overdose interval.

Value

The DLT probability vector is computed based on the MCMC samples of BLRM model parameters. This can be seen as a MCMC samples of the DLT probability. For the returning results, the samples that fails to compute the DLT probability are excluded from the summary and the number of such samples are returned also.

This function returns a named numerical vector containing the following component:

  • under: proportions in MCMC samples that the resulting DLT prob falls in underdose interval.

  • target: proportions in MCMC samples that the resulting DLT probability falls in target dose interval.

  • over: proportions in MCMC samples that the resulting DLT probability falls in overdose interval.

  • mean: mean value of DLT probabilities for this dose.

  • sd: sd of the resulting DLT probabilities samples.

  • p2.5, p50, p97.5: 2.5%, 50%, 97.5% percentile in the DLT probability samples.

  • na_num: number of MCMC samples that fails to compute the DLT probability.

Details

This function assumes the BLRM model.

$$ logit(p) = a0 + a1 * log(dose / ref\_dose) $$

a0 and a1 are sampled by crmPack::mcmc() from a crmPack::LogisticLogNormal() model. Then given this MCMC sample, we can compute the corresponding DLT probability samples.