Independent test for wlr.R is added

tests/testthat/test-independent_test_wlr.R

@@ -0,0 +1,245 @@
+#### unstratified, FH (Fleming-Harrington) ----
+# Check value when Fleming-Harrington weight is used
+test_that("wlr() with FH weight on unstratified data", {
+  # Example 1: Unstratified
+  set.seed(123456)
+  base <- sim_pw_surv(n = 200) |>
+    cut_data_by_event(125) #|>
+  output <- base |>
+    wlr(weight = fh(rho = c(0, 0, 1, 1), gamma = c(0, 1, 0, 1)))
+
+  observed <- output$z
+  base <- base |> counting_process(arm = "experimental")
+  expected <- c()
+  for (i in 1:length(observed)) {
+    base <- base |> mutate(weight=s^(output$rho[i])*(1-s)^(output$gamma[i]))
+    z <- sum(base$o_minus_e*base$weight)/sqrt(sum(base$weight^2*base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+
+
+#### stratified, FH (Fleming-Harrington) ----
+# Check value when Fleming-Harrington weight is used
+test_that("wlr() with FH weight on stratified data", {
+  # Example 1: Stratified
+  set.seed(123456)
+  n <- 500
+  # Two strata
+  stratum <- c("Biomarker-positive", "Biomarker-negative")
+  prevalence_ratio <- c(0.6, 0.4)
+  enroll_rate <- gsDesign2::define_enroll_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(2, 10, 2, 10),
+    rate = c(c(1, 4) * prevalence_ratio[1], c(1, 4) * prevalence_ratio[2])
+  )
+  enroll_rate$rate <- enroll_rate$rate * n / sum(enroll_rate$duration * enroll_rate$rate)  #??
+  # Failure rate
+  med_pos <- 10 # Median of the biomarker positive population
+  med_neg <- 8 # Median of the biomarker negative population
+  hr_pos <- c(1, 0.7) # Hazard ratio of the biomarker positive population
+  hr_neg <- c(1, 0.8) # Hazard ratio of the biomarker negative population
+  fail_rate <- gsDesign2::define_fail_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(3, 1000, 4, 1000),
+    fail_rate = c(log(2) / c(med_pos, med_pos, med_neg, med_neg)),
+    hr = c(hr_pos, hr_neg),
+    dropout_rate = 0.01
+  )
+  temp <- to_sim_pw_surv(fail_rate) # Convert the failure rate
+  set.seed(123456)
+  base <- sim_pw_surv(
+    n = n, # Sample size
+    # Stratified design with prevalence ratio of 6:4
+    stratum = tibble(stratum = stratum, p = prevalence_ratio),
+    # Randomization ratio
+    block = c("control", "control", "experimental", "experimental"),
+    enroll_rate = enroll_rate, # Enrollment rate
+    fail_rate = temp$fail_rate, # Failure rate
+    dropout_rate = temp$dropout_rate # Dropout rate
+  ) |>
+    cut_data_by_event(125)
+
+  output <- base |>
+    wlr(weight = fh(rho = c(0, 0, 1, 1), gamma = c(0, 1, 0, 1)))
+
+  observed <- output$z
+  base <- base |> counting_process(arm = "experimental")
+  expected <- c()
+  for (i in 1:length(observed)) {
+    base <- base |> mutate(weight=s^(output$rho[i])*(1-s)^(output$gamma[i]))
+    z <- sum(base$o_minus_e*base$weight)/sqrt(sum(base$weight^2*base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+
+
+#### unstratified, MB (Magirr and Burman) ----
+# Check value when Magirr and Burman weight is used
+test_that("wlr() with MB weight on unstratified data", {
+  # Example 1: Unstratified
+  set.seed(123456)
+  delay <- 4
+  w_max <- 2
+  base <- sim_pw_surv(n = 200) |>
+    cut_data_by_event(125)
+  output <- base |>
+    wlr(weight = mb(delay = delay, w_max = w_max))
+
+  observed <- output$z
+  base <- base |> counting_process(arm = "experimental")
+  base2 <- base |> filter(tte<=delay)
+  expected <- c()
+  for (i in 1:length(observed)) {
+    wht <- base2 |> group_by(stratum) %>% summarise(mx = max(1/s)) |> mutate(mx = pmin(mx,w_max))
+    base <- base |> full_join(wht, by=c('stratum')) |> mutate(weight=pmin(1/s,mx))
+    z <- sum(base$o_minus_e*base$weight)/sqrt(sum(base$weight^2*base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+
+
+#### stratified, MB (Magirr and Burman) ----
+# Check value when Magirr and Burman weight is used
+test_that("wlr() with MB weight on stratified data", {
+  # Example 2: Stratified
+  set.seed(123456)
+  delay <- 4
+  w_max <- 2
+  n <- 500
+  # Two strata
+  stratum <- c("Biomarker-positive", "Biomarker-negative")
+  prevalence_ratio <- c(0.6, 0.4)
+  enroll_rate <- gsDesign2::define_enroll_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(2, 10, 2, 10),
+    rate = c(c(1, 4) * prevalence_ratio[1], c(1, 4) * prevalence_ratio[2])
+  )
+  enroll_rate$rate <- enroll_rate$rate * n / sum(enroll_rate$duration * enroll_rate$rate)  #??
+  # Failure rate
+  med_pos <- 10 # Median of the biomarker positive population
+  med_neg <- 8 # Median of the biomarker negative population
+  hr_pos <- c(1, 0.7) # Hazard ratio of the biomarker positive population
+  hr_neg <- c(1, 0.8) # Hazard ratio of the biomarker negative population
+  fail_rate <- gsDesign2::define_fail_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(3, 1000, 4, 1000),
+    fail_rate = c(log(2) / c(med_pos, med_pos, med_neg, med_neg)),
+    hr = c(hr_pos, hr_neg),
+    dropout_rate = 0.01
+  )
+  temp <- to_sim_pw_surv(fail_rate) # Convert the failure rate
+  set.seed(123456)
+  base <- sim_pw_surv(
+    n = n, # Sample size
+    # Stratified design with prevalence ratio of 6:4
+    stratum = tibble(stratum = stratum, p = prevalence_ratio),
+    # Randomization ratio
+    block = c("control", "control", "experimental", "experimental"),
+    enroll_rate = enroll_rate, # Enrollment rate
+    fail_rate = temp$fail_rate, # Failure rate
+    dropout_rate = temp$dropout_rate # Dropout rate
+  ) |>
+    cut_data_by_event(125)
+
+  output <- base |>
+    wlr(weight = mb(delay = delay, w_max = w_max))
+
+  observed <- output$z
+  base <- base |> counting_process(arm = "experimental")
+  base2 <- base |> filter(tte<=delay)
+  expected <- c()
+  for (i in 1:length(observed)) {
+    wht <- base2 |> group_by(stratum) %>% summarise(mx = max(1/s)) |> mutate(mx = pmin(mx,w_max))
+    base <- base |> full_join(wht, by=c('stratum')) |> mutate(weight=pmin(1/s,mx))
+    z <- sum(base$o_minus_e*base$weight)/sqrt(sum(base$weight^2*base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+
+
+#### unstratified, early_zero_weight ----
+# Check value when early_zero_weight is used
+test_that("wlr() with early_zero_weight on unstratified data", {
+  # Example 1: Unstratified
+  set.seed(123456)
+  early_period = 4
+  base <- sim_pw_surv(n = 200) |>
+    cut_data_by_event(125)
+  output <- base |>
+    wlr(weight = early_zero(early_period = early_period))
+
+  observed <- output$z
+  # WLR using early_zero_weight yields the same results as directly removing the events happening earlier than `early_period`
+  base <- base |> counting_process(arm = "experimental") %>% filter(tte>=early_period)
+  expected <- c()
+  for (i in 1:length(observed)) {
+    # base <- base |> mutate(weight=if_else(tte<early_period,0,1))
+    z <- sum(base$o_minus_e)/sqrt(sum(base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+
+
+#### stratified, early_zero_weight ----
+# Check value when early_zero_weight is used
+test_that("wlr() with early_zero_weight on stratified data", {
+  # Example 2: Stratified
+  set.seed(123456)
+  early_period = 4
+  n <- 500
+  # Two strata
+  stratum <- c("Biomarker-positive", "Biomarker-negative")
+  prevalence_ratio <- c(0.6, 0.4)
+  enroll_rate <- gsDesign2::define_enroll_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(2, 10, 2, 10),
+    rate = c(c(1, 4) * prevalence_ratio[1], c(1, 4) * prevalence_ratio[2])
+  )
+  enroll_rate$rate <- enroll_rate$rate * n / sum(enroll_rate$duration * enroll_rate$rate)  #??
+  # Failure rate
+  med_pos <- 10 # Median of the biomarker positive population
+  med_neg <- 8 # Median of the biomarker negative population
+  hr_pos <- c(1, 0.7) # Hazard ratio of the biomarker positive population
+  hr_neg <- c(1, 0.8) # Hazard ratio of the biomarker negative population
+  fail_rate <- gsDesign2::define_fail_rate(
+    stratum = rep(stratum, each = 2),
+    duration = c(3, 1000, 4, 1000),
+    fail_rate = c(log(2) / c(med_pos, med_pos, med_neg, med_neg)),
+    hr = c(hr_pos, hr_neg),
+    dropout_rate = 0.01
+  )
+  temp <- to_sim_pw_surv(fail_rate) # Convert the failure rate
+  set.seed(123456)
+  x <- sim_pw_surv(
+    n = n, # Sample size
+    # Stratified design with prevalence ratio of 6:4
+    stratum = tibble(stratum = stratum, p = prevalence_ratio),
+    # Randomization ratio
+    block = c("control", "control", "experimental", "experimental"),
+    enroll_rate = enroll_rate, # Enrollment rate
+    fail_rate = temp$fail_rate, # Failure rate
+    dropout_rate = temp$dropout_rate # Dropout rate
+  ) |>
+    cut_data_by_event(125)
+
+  output <- base |>
+    wlr(weight = early_zero(early_period = early_period))
+
+  observed <- output$z
+  # WLR using early_zero_weight yields the same results as directly removing the events happening earlier than `early_period`
+  base <- base |> counting_process(arm = "experimental") %>% filter(tte>=early_period)
+  expected <- c()
+  for (i in 1:length(observed)) {
+    # base <- base |> mutate(weight=if_else(tte<early_period,0,1))
+    z <- sum(base$o_minus_e)/sqrt(sum(base$var_o_minus_e))
+    expected <- c(expected,z)
+  }
+  expect_equal(observed, expected)
+})
+