BayesianHybridDesign (version 0.1.0)

Bayesian.Hybrid.Analysis: Statistical Analysis for a Bayesian Hybrid Design

Description

This function performs the statistical analysis for a Bayesian Hybrid Design using a dynamic power prior approach.

Usage

Bayesian.Hybrid.Analysis(
  Yt = 20,
  nt = 40,
  Yc = 12,
  nc = 40,
  Ych = 73,
  nche = 40,
  nch = 234,
  sig = 0.9,
  credlev = 0.8,
  a0c = 0.001,
  b0c = 0.001,
  a0t = 0.001,
  b0t = 0.001,
  delta_threshold = 0.1
)

Value

A list containing the following components:

• prob.pt.gt.pc: Probability of experimental arm having a better posterior response rate than control.

• median_hca: Posterior median response rate for hybrid control.

• CI_hca: Credible interval for median response rate for hybrid control.

• median_c: Posterior median response rate for current study control.

• CI_c: Credible interval for median response rate for current study control.

• median_t: Posterior median response rate for current study experimental arm.

• CI_t: Credible interval for median response rate for current experimental arm.

• delta.m: Posterior median response rate difference (experimental - control) based on hybrid design.

• delta.CI: Credible interval for response rate difference based on hybrid design.

• delta.m_trial: Posterior median response rate difference (experimental - control) based on current study only.

• delta.CI_trial: Credible interval for response rate difference based on current study only.

• conclusion: Statistical inference conclusion string.

Arguments

Yt

Number of responses in the experimental arm in the current study.

nt

Number of patients in the experimental arm in the current study.

Yc

Number of responses in the control arm in the current study.

nc

Number of patients in the control arm in the current study.

Ych

Number of responses in the control treatment in the historical study.

nche

Equivalent number of patients borrowed from the historical study.

nch

Total number of patients in the historical control.

sig

Significance boundary. The hypothesis is considered significant if the posterior probability \(P(p_t > p_c | data) > sig\).

credlev

Credible interval level (e.g., 0.95 for 95 percent CI).

a0c

Prior alpha parameter for control response rate, \(Beta(a_{0c}, b_{0c})\).

b0c

Prior beta parameter for control response rate, \(Beta(a_{0c}, b_{0c})\).

a0t

Prior alpha parameter for experimental response rate, \(Beta(a_{0t}, b_{0t})\).

b0t

Prior beta parameter for experimental response rate, \(Beta(a_{0t}, b_{0t})\).

delta_threshold

Borrowing threshold. Borrowing occurs when \(|p_{c,trial} - p_{c,hist}| \le\) delta_threshold.

Examples

# \donttest{
# Note: This example relies on the internal package function 'borrow.wt'
Bayesian.Hybrid.Analysis(Yt=18, nt=40, Yc=13, nc=40, Ych=73,
                         nche=40, nch=234)
# }