validation/test_code/01_BasicSettingTestCode.R
In Sterniii3/drugdevelopR: Utility-Based Optimal Phase II/III Drug Development Planning
#' @editor Lukas D Sauer
#' @editDate 2022-09-13
test_that("01.01", {
# Testing time to event with prior distribution
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
expect_equal(res$n2, 206) # optimal sample size in phase II
expect_equal(res$n3, 354) # resulting sample size in phase III
expect_equal(res$n, 560) # resulting total sample size
expect_equal(res$u, 432, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.84) # threshold for proceeding to phase III
expect_equal(res$d2, 144) # expected number of events in phase II
expect_equal(res$d3, 248) # expected number of events in phase III
expect_equal(res$d, 392) # total expected number of events
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.02", {
# Time to event with fixed treatment effects
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.8, hr2 = NULL, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = TRUE, # use a prior distribution
w = NULL, # weight for the prior distribution
id1 = NULL, id2 = NULL, # amount of information (number of events) for prior
# true treatment effect
)
expect_equal(res$n2, 240) # optimal sample size in phase II
expect_equal(res$u, 352, tolerance = 0.001) # expected utility
expect_equal(res$HRgo, 0.88) # threshold for proceeding to phase III
expect_equal(res$pgo, 0.73) # probability to proceed to phase III
expect_equal(res$d2, 168) # expected number of events in phase II
expect_equal(res$d3, 546) # expected number of events in phase III
expect_equal(res$d, 714) # total expected number of events
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.03", {
# Testing time to event with cost constraints
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
K = 750 # cost constraint
)
expect_equal(res$n2, 228) # optimal sample size in phase II
expect_equal(res$u, 996, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.84)
expect_equal(res$K2, 271)
expect_equal(res$K3, 478)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.04", {
# Testing time to event with sample size constraint
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
N = 500 # sample size constraint
)
expect_equal(res$n2, 170) # optimal sample size in phase II
expect_equal(res$n3, 328) # resulting sample size in phase III
expect_equal(res$n, 498) # resulting total sample size
expect_equal(res$u, 956, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.83) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.05", {
# Testing time to event with success probability constraint
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
S = 0.6 # constraint on success probability
)
expect_equal(res$n2, 470) # optimal sample size in phase II
expect_equal(res$u, 899, tolerance = 0.01) # expected utility
expect_equal(res$HRgo, 0.89) # threshold for proceeding to phase III
expect_equal(res$pgo, 0.77) # probability to proceed to phase III
expect_equal(res$sProg, 0.6) # probability of a successful program
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.06", {
# Testing time to event while to skipping phase II
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
skipII = TRUE # skip phase II
)
expect_equal(res[2,]$n3, 824) # optimal sample size in phase III after skipping phase II
expect_equal(res[2,]$u, 1706, tolerance = 0.0001) # expected utility
expect_equal(res[2,]$HRgo, Inf) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.07", {
# Testing time to event while modeling differing population structures in phase II and III
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
gamma = 0.025
)
expect_equal(res$n2, 310) # optimal sample size in phase II
expect_equal(res$u, 1207, tolerance = 0.0005) # expected utility
expect_equal(res$HRgo, 0.86) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-12
test_that("01.08", {
# Testing binary endpoints with fixed effects
res = optimal_binary(alpha = 0.025,
beta = 0.1,
p0 = 0.6, p11 = 0.5, p12 = NULL,
n2min = 10, n2max = 500, stepn2 = 2,
rrgomin = 0.7, rrgomax = 0.9, steprrgo = 0.01,
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85,
b1 = 1000, b2 = 3000, b3 = 5000,
num_cl = 12,
c02 = 100, c03 = 150,
c2 = 0.75, c3 = 1,
fixed = TRUE,
w = NULL,
in1 = NULL, in2 = NULL
)
expect_equal(res$n2, 204)
expect_equal(res$u, 299, tolerance = 0.0005)
expect_equal(res$RRgo, 0.90)
expect_equal(res$K, Inf)
expect_equal(res$K2, 253)
expect_equal(res$K3, 769)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-12
test_that("01.09", {
# Testing binary endpoints with treatment effects modelled on a prior distribution
res = optimal_binary(alpha = 0.025,
beta = 0.1,
p0 = 0.6, p11 = 0.3, p12 = 0.5,
n2min = 10, n2max = 500, stepn2 = 2,
rrgomin = 0.7, rrgomax = 0.9, steprrgo = 0.01,
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85,
b1 = 1000, b2 = 3000, b3 = 5000,
num_cl = 12,
c02 = 100, c03 = 150,
c2 = 0.75, c3 = 1,
fixed = FALSE,
w = 0.4,
in1 = 30, in2 = 60
)
expect_equal(res$n2, 224)
expect_equal(res$u, 1542, tolerance = 0.0005)
expect_equal(res$RRgo, 0.89)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.10", {
# Testing normally distributed endpoints with prior distribution on effects
res = optimal_normal(
alpha = 0.025,
beta = 0.1,
Delta1 = 0.375, Delta2 = 0.5,
n2min = 10, n2max = 500, stepn2 = 2,
kappamin = 0.01, kappamax = 0.5, stepkappa = 0.01,
steps1 = 0, stepm1 = 0.375, stepl1 = 0.625,
b1 = 625, b2 = 2000, b3 = 10000,
num_cl = 12,
c02 = 15, c03 = 20,
c2 = 0.675, c3 = 0.72,
fixed = FALSE,
w = 0.5,
in1 = 300, in2 = 600,
a = 0, b = 0.75
)
expect_equal(res$n2, 86) # optimal sample size in phase II
expect_equal(res$u, 337, tolerance = 0.001) # expected utility
expect_equal(res$Kappa, 0.19) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.11", {
# Testing normally distributed endpoints with fixed effects
res = optimal_normal(
alpha = 0.025,
beta = 0.1,
Delta1 = 0.625, Delta2 = NULL,
n2min = 10, n2max = 500, stepn2 = 2,
kappamin = 0.01, kappamax = 0.5, stepkappa = 0.01,
steps1 = 0, stepm1 = 0.375, stepl1 = 0.625,
b1 = 625, b2 = 2000, b3 = 10000,
num_cl = 12,
c02 = 15, c03 = 20,
c2 = 0.675, c3 = 0.72,
fixed = TRUE,
w = NULL,
# 300 events in phase II and 600 events in phase III, Wo finde ich diese Parameter?
# Vielleicht ist folgendes gemeint?
in1 = NULL, in2 = NULL,
a = NULL, b = NULL
)
expect_equal(res$n2, 78) # optimal sample size in phase II
expect_equal(res$u, 944, tolerance = 0.0001) # expected utility
expect_equal(res$Kappa, 0.12) # threshold for proceeding to phase III
expect_equal(res$sProg, 0.83) # probability of a successful program
expect_equal(res$sProg1, 0.51) # probability of a successful program with small treatment effect
expect_equal(res$sProg2, 0.30) # probability of a successful program with medium treatment effect
expect_equal(res$sProg3, 0.02) # probability of a successful program with large treatment effect
}
)
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.12", {
# Testing that parallel computing has an effect
start_time_3 = Sys.time()
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
end_time_3 = Sys.time()
time_elapsed_01_02_num_cl_3 = end_time_3 - start_time_3
start_time_1 = Sys.time()
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 1, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
end_time_1 = Sys.time()
time_elapsed_01_02_num_cl_1 = end_time_1 - start_time_1
expect_true(time_elapsed_01_02_num_cl_1 > time_elapsed_01_02_num_cl_3)
}
)
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.13", {
# Comparing results of Epgo_normal() and En3_normal() to SAS results
data <- haven::read_sas("../ref/valref_normal.sas7bdat")
Epgo_r <- rep(0, nrow(data))
En3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_normal(kappa = data$kappa_vec[i], n2 = data$n2_vec[i],
Delta1 = data$Delta1_vec[i], Delta2 = NULL,
w = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
En3_r[i] <- drugdevelopR:::En3_normal(kappa = data$kappa_vec[i], n2 = data$n2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
Delta1 = data$Delta1_vec[i], Delta2 = NULL,
w = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(En3_r, data$n3_sas)
})
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.14", {
# Comparing results of Epgo_binary() and En3_binary() to SAS results
data <- haven::read_sas("../ref/valref_binary.sas7bdat")
Epgo_r <- rep(0, nrow(data))
En3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_binary(RRgo = data$RRgo_vec[i], n2 = data$n2_vec[i],
p0 = data$p0_vec[i], p11 = data$p11_vec[i],
p12 = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
En3_r[i] <- drugdevelopR:::En3_binary(RRgo = data$RRgo_vec[i], n2 = data$n2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
p0 = data$p0_vec[i], w = NULL, p11 = data$p11_vec[i],
p12 = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(En3_r, data$n3_sas)
})
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.15", {
# Comparing results of Epgo_tte() and En3_tte() to SAS results
data <- haven::read_sas("../ref/valref_tte.sas7bdat")
Epgo_r <- rep(0, nrow(data))
Ed3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_tte(HRgo = data$HRgo_vec[i], d2 = data$d2_vec[i],
hr1 = data$hr1_vec[i], hr2 = NULL,
w = NULL, id1 = NULL, id2 = NULL, fixed = TRUE)
Ed3_r[i] <- drugdevelopR:::Ed3_tte(HRgo = data$HRgo_vec[i], d2 = data$d2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
hr1 = data$hr1_vec[i], hr2 = NULL,
w = NULL, id1 = NULL, id2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(Ed3_r, data$d3_sas)
})
Sterniii3/drugdevelopR documentation built on Jan. 26, 2024, 6:17 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#' @editor Lukas D Sauer
#' @editDate 2022-09-13
test_that("01.01", {
# Testing time to event with prior distribution
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
expect_equal(res$n2, 206) # optimal sample size in phase II
expect_equal(res$n3, 354) # resulting sample size in phase III
expect_equal(res$n, 560) # resulting total sample size
expect_equal(res$u, 432, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.84) # threshold for proceeding to phase III
expect_equal(res$d2, 144) # expected number of events in phase II
expect_equal(res$d3, 248) # expected number of events in phase III
expect_equal(res$d, 392) # total expected number of events
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.02", {
# Time to event with fixed treatment effects
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.8, hr2 = NULL, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = TRUE, # use a prior distribution
w = NULL, # weight for the prior distribution
id1 = NULL, id2 = NULL, # amount of information (number of events) for prior
# true treatment effect
)
expect_equal(res$n2, 240) # optimal sample size in phase II
expect_equal(res$u, 352, tolerance = 0.001) # expected utility
expect_equal(res$HRgo, 0.88) # threshold for proceeding to phase III
expect_equal(res$pgo, 0.73) # probability to proceed to phase III
expect_equal(res$d2, 168) # expected number of events in phase II
expect_equal(res$d3, 546) # expected number of events in phase III
expect_equal(res$d, 714) # total expected number of events
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.03", {
# Testing time to event with cost constraints
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
K = 750 # cost constraint
)
expect_equal(res$n2, 228) # optimal sample size in phase II
expect_equal(res$u, 996, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.84)
expect_equal(res$K2, 271)
expect_equal(res$K3, 478)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.04", {
# Testing time to event with sample size constraint
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
N = 500 # sample size constraint
)
expect_equal(res$n2, 170) # optimal sample size in phase II
expect_equal(res$n3, 328) # resulting sample size in phase III
expect_equal(res$n, 498) # resulting total sample size
expect_equal(res$u, 956, tolerance = 0.05) # expected utility
expect_equal(res$HRgo, 0.83) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.05", {
# Testing time to event with success probability constraint
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
S = 0.6 # constraint on success probability
)
expect_equal(res$n2, 470) # optimal sample size in phase II
expect_equal(res$u, 899, tolerance = 0.01) # expected utility
expect_equal(res$HRgo, 0.89) # threshold for proceeding to phase III
expect_equal(res$pgo, 0.77) # probability to proceed to phase III
expect_equal(res$sProg, 0.6) # probability of a successful program
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-10
test_that("01.06", {
# Testing time to event while to skipping phase II
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
skipII = TRUE # skip phase II
)
expect_equal(res[2,]$n3, 824) # optimal sample size in phase III after skipping phase II
expect_equal(res[2,]$u, 1706, tolerance = 0.0001) # expected utility
expect_equal(res[2,]$HRgo, Inf) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.07", {
# Testing time to event while modeling differing population structures in phase II and III
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.6, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
gamma = 0.025
)
expect_equal(res$n2, 310) # optimal sample size in phase II
expect_equal(res$u, 1207, tolerance = 0.0005) # expected utility
expect_equal(res$HRgo, 0.86) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-12
test_that("01.08", {
# Testing binary endpoints with fixed effects
res = optimal_binary(alpha = 0.025,
beta = 0.1,
p0 = 0.6, p11 = 0.5, p12 = NULL,
n2min = 10, n2max = 500, stepn2 = 2,
rrgomin = 0.7, rrgomax = 0.9, steprrgo = 0.01,
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85,
b1 = 1000, b2 = 3000, b3 = 5000,
num_cl = 12,
c02 = 100, c03 = 150,
c2 = 0.75, c3 = 1,
fixed = TRUE,
w = NULL,
in1 = NULL, in2 = NULL
)
expect_equal(res$n2, 204)
expect_equal(res$u, 299, tolerance = 0.0005)
expect_equal(res$RRgo, 0.90)
expect_equal(res$K, Inf)
expect_equal(res$K2, 253)
expect_equal(res$K3, 769)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-12
test_that("01.09", {
# Testing binary endpoints with treatment effects modelled on a prior distribution
res = optimal_binary(alpha = 0.025,
beta = 0.1,
p0 = 0.6, p11 = 0.3, p12 = 0.5,
n2min = 10, n2max = 500, stepn2 = 2,
rrgomin = 0.7, rrgomax = 0.9, steprrgo = 0.01,
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85,
b1 = 1000, b2 = 3000, b3 = 5000,
num_cl = 12,
c02 = 100, c03 = 150,
c2 = 0.75, c3 = 1,
fixed = FALSE,
w = 0.4,
in1 = 30, in2 = 60
)
expect_equal(res$n2, 224)
expect_equal(res$u, 1542, tolerance = 0.0005)
expect_equal(res$RRgo, 0.89)
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.10", {
# Testing normally distributed endpoints with prior distribution on effects
res = optimal_normal(
alpha = 0.025,
beta = 0.1,
Delta1 = 0.375, Delta2 = 0.5,
n2min = 10, n2max = 500, stepn2 = 2,
kappamin = 0.01, kappamax = 0.5, stepkappa = 0.01,
steps1 = 0, stepm1 = 0.375, stepl1 = 0.625,
b1 = 625, b2 = 2000, b3 = 10000,
num_cl = 12,
c02 = 15, c03 = 20,
c2 = 0.675, c3 = 0.72,
fixed = FALSE,
w = 0.5,
in1 = 300, in2 = 600,
a = 0, b = 0.75
)
expect_equal(res$n2, 86) # optimal sample size in phase II
expect_equal(res$u, 337, tolerance = 0.001) # expected utility
expect_equal(res$Kappa, 0.19) # threshold for proceeding to phase III
})
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.11", {
# Testing normally distributed endpoints with fixed effects
res = optimal_normal(
alpha = 0.025,
beta = 0.1,
Delta1 = 0.625, Delta2 = NULL,
n2min = 10, n2max = 500, stepn2 = 2,
kappamin = 0.01, kappamax = 0.5, stepkappa = 0.01,
steps1 = 0, stepm1 = 0.375, stepl1 = 0.625,
b1 = 625, b2 = 2000, b3 = 10000,
num_cl = 12,
c02 = 15, c03 = 20,
c2 = 0.675, c3 = 0.72,
fixed = TRUE,
w = NULL,
# 300 events in phase II and 600 events in phase III, Wo finde ich diese Parameter?
# Vielleicht ist folgendes gemeint?
in1 = NULL, in2 = NULL,
a = NULL, b = NULL
)
expect_equal(res$n2, 78) # optimal sample size in phase II
expect_equal(res$u, 944, tolerance = 0.0001) # expected utility
expect_equal(res$Kappa, 0.12) # threshold for proceeding to phase III
expect_equal(res$sProg, 0.83) # probability of a successful program
expect_equal(res$sProg1, 0.51) # probability of a successful program with small treatment effect
expect_equal(res$sProg2, 0.30) # probability of a successful program with medium treatment effect
expect_equal(res$sProg3, 0.02) # probability of a successful program with large treatment effect
}
)
#' @editor Lukas D Sauer
#' @editDate 2022-10-11
test_that("01.12", {
# Testing that parallel computing has an effect
start_time_3 = Sys.time()
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 12, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
end_time_3 = Sys.time()
time_elapsed_01_02_num_cl_3 = end_time_3 - start_time_3
start_time_1 = Sys.time()
res = optimal_tte(
alpha = 0.025, # significance level
beta = 0.1, # 1 - power
hr1 = 0.69, hr2 = 0.88, # assumed treatment effects
xi2 = 0.7, xi3 = 0.7, # event rates
d2min = 10, d2max = 400, stepd2 = 1, # optimization region for the number of events
hrgomin = 0.71, hrgomax = 0.95, stephrgo = 0.01, # optimization
# region for the threshold values
steps1 = 1, stepm1 = 0.95, stepl1 = 0.85, # boundaries for the effect size
# categories small, medium and large
b1 = 1000, b2 = 3000, b3 = 5000, # expected gains for each effect size
num_cl = 1, # number of clusters
c02 = 100, c03 = 150, # fixed cost for phase II and phase III
c2 = 0.75, c3 = 1, # variable per-patient cost in phase II and phase III
fixed = FALSE, # use a prior distribution
w = 0.3, # weight for the prior distribution
id1 = 210, id2 = 420, # amount of information (number of events) for prior
# true treatment effect
)
end_time_1 = Sys.time()
time_elapsed_01_02_num_cl_1 = end_time_1 - start_time_1
expect_true(time_elapsed_01_02_num_cl_1 > time_elapsed_01_02_num_cl_3)
}
)
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.13", {
# Comparing results of Epgo_normal() and En3_normal() to SAS results
data <- haven::read_sas("../ref/valref_normal.sas7bdat")
Epgo_r <- rep(0, nrow(data))
En3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_normal(kappa = data$kappa_vec[i], n2 = data$n2_vec[i],
Delta1 = data$Delta1_vec[i], Delta2 = NULL,
w = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
En3_r[i] <- drugdevelopR:::En3_normal(kappa = data$kappa_vec[i], n2 = data$n2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
Delta1 = data$Delta1_vec[i], Delta2 = NULL,
w = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(En3_r, data$n3_sas)
})
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.14", {
# Comparing results of Epgo_binary() and En3_binary() to SAS results
data <- haven::read_sas("../ref/valref_binary.sas7bdat")
Epgo_r <- rep(0, nrow(data))
En3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_binary(RRgo = data$RRgo_vec[i], n2 = data$n2_vec[i],
p0 = data$p0_vec[i], p11 = data$p11_vec[i],
p12 = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
En3_r[i] <- drugdevelopR:::En3_binary(RRgo = data$RRgo_vec[i], n2 = data$n2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
p0 = data$p0_vec[i], w = NULL, p11 = data$p11_vec[i],
p12 = NULL, in1 = NULL, in2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(En3_r, data$n3_sas)
})
#' @editor Lukas D Sauer
#' @editDate 2023-03-30
test_that("01.15", {
# Comparing results of Epgo_tte() and En3_tte() to SAS results
data <- haven::read_sas("../ref/valref_tte.sas7bdat")
Epgo_r <- rep(0, nrow(data))
Ed3_r <- rep(0, nrow(data))
for(i in (1:nrow(data))){
Epgo_r[i] <- drugdevelopR:::Epgo_tte(HRgo = data$HRgo_vec[i], d2 = data$d2_vec[i],
hr1 = data$hr1_vec[i], hr2 = NULL,
w = NULL, id1 = NULL, id2 = NULL, fixed = TRUE)
Ed3_r[i] <- drugdevelopR:::Ed3_tte(HRgo = data$HRgo_vec[i], d2 = data$d2_vec[i],
alpha = data$alpha_vec[i], beta = data$beta_vec[i],
hr1 = data$hr1_vec[i], hr2 = NULL,
w = NULL, id1 = NULL, id2 = NULL, fixed = TRUE)
}
expect_equal(Epgo_r, data$pgo_sas)
expect_equal(Ed3_r, data$d3_sas)
})
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.