response: Random response generator

In yevgenryeznik/artool: Adaptive Randomization Routine for Clinical Trials

Description

Generates random efficacy responses of subjects randomized in a clinical trial

Usage

response(distr, distr_param, treatment)

Arguments

distr	distribution of responses. The following values are supported: "binary" – Binary (Bernoulli) distribution, "uniform" – Uniform distribution, "normal" – Normal distribution, "exponential" – Exponential distribution, "weibull" – Weibull distribution, "loglogistic" – Log-logistic distribution, "lognormal" – Log-normal distribution.
distr_param	parameter(s) of the response distribution (a named list with numeric vector(s) of length K each, where K is a total number of treatments studied in a clinical trial. It must have the following structure depending on a value of the distr parameter: distr_param = list(prob) if distr == "binary"; in this case response is a binary variable with values {0, 1}. It has a parameter prob = p, and Pr(response = 1) = p, distr_param = list(min, max) if distr == "uniform"; the Unifrom distribution has a p.d.f. f(x) = 1/(max-min), distr_param = list(mean, sd) "normal"; it is assumed that a p.d.f. of Normal distribution with mean = μ and sd = σ equals to f(x) = 1/(σ√(2π))exp(-(x-μ)^2/(2σ^2)), distr_param = list(rate) if distr == "exponential"; it is assumed that a p.d.f. of Exponential distribution with rate = λ equals to f(x) = λ exp(-λ x), distr_param = list(shape, scale) if distr == "weibull"; it is assumed that a p.d.f. of Weibull distribution with shape = k and scale = λ equals to f(x) = (k/λ)(x/k)^(k-1)exp(-(x/λ)^k), distr_param = list(shape, scale) if distr == "loglogistic"; it is assumed that a p.d.f. of Log-logistic distribution with shape = β and scale = α equals to f(x) = (β/α)(x/α)^(β-1)/(1+(x/α)^β)^2, distr_param = list(mu, sigma) "lognormal"; it is assumed that a p.d.f. of Log-normal distribution with mu = μ and sigma = σ equals to f(x) = 1/(xσ√(2π))exp(-(lnx-μ)^2/(2σ^2)).
treatment	vector of treatment assignments (contains integers from 1 to K, where K is a total number of treatments studied in a clinical trial).

Value

a numeric vector of rsponses' values

Author(s)

Yevgen Ryeznik ([e-mail]), Oleksandr Sverdlov

Examples

   library(artool)
   
   # generate random treatment assignments (4 treatments, equal probabilities)
   treatment <- sample(1:4, 100, replace = TRUE, prob = rep(0.25,4))
   
   # generate normal responses with the same means = 0 and all sds = 1
   response("normal", list(mean = rep(0,4), sd = rep(1, 4)), treatment)
   
   # generate normal responses with means c(0, 0.2, 0.4, 0.6), and sd's = c(0.1, 0.5, 0.7, 1)
   response("normal", list(mean = c(0, 0.2, 0.4, 0.6), sd = c(0.1, 0.5, 0.7, 1)), treatment)
   
   # generate Weibul responses with shapes = c(0.5, 1, 1.5, 2) and scale = 1
   response("weibull", list(shape = c(0.5, 1, 1.5, 2), scale = rep(1, 4)), treatment)
    

yevgenryeznik/artool documentation built on May 15, 2019, 5:35 p.m.