tests/testit/test-independent-gs_info_ahr.R

In gsDesign2: Group Sequential Design with Non-Constant Effect

# Test 1: independent test using AHR to check outputs of gs_info_ahr

assert("results match if only put in targeted analysis times", {
  res <- test_gs_info_ahr()
  enroll_rate <- res$enroll_rate
  fail_rate <- res$fail_rate
  total_duration <- c(18, 27, 36)

  res <- gs_info_ahr(
    enroll_rate = enroll_rate,
    fail_rate = fail_rate,
    analysis_time = total_duration
  ) |> dplyr::select(time, ahr, event, info, info0)
  expected <- ahr(
    enroll_rate = enroll_rate,
    fail_rate = fail_rate,
    total_duration = total_duration
  ) |> dplyr::select(-n)
  (res %==% expected)
})

assert("results match if only put in targeted events", {
  res <- test_gs_info_ahr()
  enroll_rate <- res$enroll_rate
  fail_rate <- res$fail_rate
  event <- c(30, 40, 50)

  out1 <- gs_info_ahr(enroll_rate = enroll_rate, fail_rate = fail_rate, event = event)

  total_duration <- out1$time

  res <- out1 |> dplyr::select(time, ahr, event, info, info0)
  expected <- ahr(
    enroll_rate = enroll_rate,
    fail_rate = fail_rate,
    total_duration = total_duration
  ) |> dplyr::select(-n)
  (res %==% expected)

  # Since above test is based on the output "time", here is to check whether
  # the output "Time" is reasonable.

  # "Time" should be at the time points when targeted event numbers are achieved
  (round(out1$event) %==% round(event))
})

assert("results match if put in both analysis time and targeted events", {
  res <- test_gs_info_ahr()
  enroll_rate <- res$enroll_rate
  fail_rate <- res$fail_rate
  event <- c(30, 40, 50)
  analysis_time <- c(16, 19, 26)

  out1 <- gs_info_ahr(
    enroll_rate = enroll_rate,
    fail_rate = fail_rate,
    event = event,
    analysis_time = analysis_time
  )

  total_duration <- out1$time

  res <- out1 |> dplyr::select(time, ahr, event, info, info0)
  expected <- ahr(
    enroll_rate = enroll_rate,
    fail_rate = fail_rate,
    total_duration = total_duration
  ) |> dplyr::select(-n)
  (res %==% expected)

  # Since above test is based on the output "Time",
  # here is to check whether the output "Time" is reasonable.

  # Either being equal to the corresponding element in the input
  # analysis_time or at the time point when targeted event number achieved.
  res <- max((1 - (out1$time == analysis_time)) * (1 - (round(out1$event) == round(event))))
  (res %==% 0)

  # "Time" >= input analysis_time
  (max(out1$time - analysis_time) >= 0)

  # "Events" >= input events
  (max(round(out1$event) - round(event)) >= 0)
})