tests/testthat/test-interpolate.conc.R

In PKNCA: Perform Pharmacokinetic Non-Compartmental Analysis

# expected errors ####

## interpolate.conc expected errors ####

test_that("interpolate.conc expected errors", {
  # Ensure that it is an error to extrapolate past the last point.
  expect_error(
    interpolate.conc(
      conc=c(0, 1, 0.5, 1, 0),
      time=0:4,
      time.out=5,
      interp.method="lin up/log down"
    ),
    regexp="`interpolate.conc()` does not extrapolate, use `interp.extrap.conc()`",
    fixed=TRUE
  )
  # Change documentation when this is not true
  expect_error(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=0:1
    ),
    regexp="Assertion on 'time.out' failed: Must have length 1."
  )
  
  expect_error(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=0.5,
      interp.method="this doesn't work"
    ),
    regexp=tryCatch(expr={
      match.arg("foo", choices=c("lin up/log down", "linear"))
    },
    error=function(e) e)$message,
    fixed=TRUE,
    info="Confirm that invalid interpolation methods are an error."
  )
  
  expect_error(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=0.5,
      conc.origin=1:2
    ),
    regexp="Assertion on 'conc.origin' failed: Must have length 1."
  )
  expect_error(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=0.5,
      conc.origin="A"
    ),
    regexp="Assertion on 'conc.origin' failed: Must be of type 'number', not 'character'."
  )
  expect_error(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=0.5,
      conc.origin=factor("A")
    ),
    regexp="Assertion on 'conc.origin' failed: Must be of type 'number', not 'factor'."
  )
  
  expect_error(
    interpolate.conc(
      conc=c(NA, 1),
      time=c(0, 1),
      time.out=0.5,
      check=FALSE
    ),
    regexp="Contains missing values"
  )

  # When tlast is before the end of the data, do not extrapolate  
  expect_error(
    interpolate.conc(
      conc=c(1, 2, 0),
      time=0:2,
      time.out=1.5
    ),
    regexp="`interpolate.conc()` can only works through Tlast, please use `interp.extrap.conc()` to combine both interpolation and extrapolation.",
    fixed=TRUE
  )
})

## extrapolate.conc expected errors ####

test_that("extrapolate.conc expected errors", {
  # Confirm that the extrap.method must be AUCinf, AUClast, AUCall
  expect_error(
    extrapolate.conc(
      conc=1,
      time=1,
      time.out=2,
      extrap.method="wrong"
    ),
    regexp="extrap.method must be one of 'AUCinf', 'AUClast', or 'AUCall'"
  )
  
  # Confirm that time.out may only be a scalar
  expect_error(
    extrapolate.conc(
      conc=1,
      time=1,
      time.out=c(2, 3),
      extrap.method="AUCinf"
    ),
    regexp="Only one time.out value may be estimated at once."
  )
  # Confirm that it is an error to extrapolate at or before Tlast
  expect_error(
    extrapolate.conc(
      conc=1,
      time=1,
      time.out=0.5,
      extrap.method="AUCinf"
    ),
    regexp="extrapolate.conc can only work beyond Tlast, please use interp.extrap.conc to combine both interpolation and extrapolation."
  )
  expect_error(
    extrapolate.conc(
      conc=1,
      time=1,
      time.out=1,
      extrap.method="AUCinf"
    ),
    regexp="extrapolate.conc can only work beyond Tlast, please use interp.extrap.conc to combine both interpolation and extrapolation."
  )
})

## interp.extrap.conc expected errors ####

test_that("interp.extrap.conc expected errors", {
  # Check inputs
  expect_error(
    interp.extrap.conc(
      conc=c(0, 1, 0.5, 1, 0),
      time=0:4,
      time.out=numeric(),
      interp.method="lin up/log down"
    ),
    regexp="time.out must be a vector with at least one element"
  )
})

## interp.extrap.conc.dose expected errors ####

test_that("interp.extrap.conc.dose expected errors", {
  # Check inputs
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose="foo",
      duration.dose=NA,
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="route.dose must be either 'extravascular' or 'intravascular'",
    info="Route must be valid"
  )
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose=factor("foo"),
      duration.dose=NA,
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="route.dose must be either 'extravascular' or 'intravascular'",
    info="Route must be valid (as a factor)"
  )
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose=c("extravascular", "extravascular"),
      duration.dose=NA,
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="route.dose must either be a scalar or the same length as time.dose",
    info="Route must have the correct length"
  )
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose="extravascular",
      duration.dose="A",
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="duration.dose must be NA or a number.",
    info="duration.dose must be NA or a number (character)."
  )
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose="extravascular",
      duration.dose=factor("A"),
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="duration.dose must be NA or a number.",
    info="duration.dose must be NA or a number (factor)."
  )
  expect_error(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      route.dose="extravascular",
      duration.dose=c(1, NA),
      time.out=c(-1, -0.1, 0, 0.1, 7),
      out.after=FALSE
    ),
    regexp="duration.dose must either be a scalar or the same length as time.dose",
    info="duration.dose must match the length of time.dose or be a scalar."
  )
})

# interpolate.conc ####

test_that("interpolate.conc", {
  # Confirm that interpolating between all zeros results in zero
  interp_method <- c("linear", "lin up/log down")
  for (n in interp_method) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 0),
        time=0:1,
        time.out=0,
        interp.method=n
      ),
      0,
      info=paste("all zero", n)
    )
  }
  
  # Confirm that interpolating to a given point results in the point itself
  interpolations <-
    list(
      linear=0,
      "lin up/log down"=0
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1),
        time=0:1,
        time.out=0,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  interpolations <-
    list(
      linear=1,
      "lin up/log down"=1
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1),
        time=0:1,
        time.out=1,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  # Confirm that interpolating to a given point when that point is NA gives a
  # interpolates as expected.
  interpolations <-
    list(
      linear=1,
      "lin up/log down"=1
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, NA, 1),
        time=0:3,
        time.out=2,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  # Slightly less trivial tests
  interpolations <-
    list(
      linear=0.5,
      "lin up/log down"=0.5
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, 0),
        time=0:2,
        time.out=0.5,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  interpolations <-
    list(
      linear=1.75,
      "lin up/log down"=2^0.75
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 2, 1),
        time=0:2,
        time.out=1.25,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  # Relative time is the only thing that matters.  Negative time
  # values are fine.
  interpolations <-
    list(
      linear=1.75,
      "lin up/log down"=2^0.75
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 2, 1),
        time=seq(-10, -8, by=1),
        time.out=-8.75,
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  # Ensure that linear-up/log down interpolation works linearly for the "up"
  # even if it is after a down.
  interpolations <-
    list(
      linear=0.25,
      "lin up/log down"=0.25
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, 0, 1, 0),
        time=0:4,
        time.out=2.25,
        conc.blq="keep",
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  # Ensure that interpolation with linear works with a BLQ in the middle: going
  # down and up from that BLQ
  interpolations <-
    list(
      linear=0.5,
      "lin up/log down"=0.5
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, 0, 1, 0),
        time=0:4,
        time.out=1.5,
        conc.blq="keep",
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  interpolations <-
    list(
      linear=0.5,
      "lin up/log down"=0.5
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, 0, 1, 0),
        time=0:4,
        time.out=2.5,
        conc.blq="keep",
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  
  # Ensure that conc.blq is appropriately observed
  interpolations <-
    list(
      linear=1,
      "lin up/log down"=1
    )
  for (n in names(interpolations)) {
    expect_equal(
      interpolate.conc(
        conc=c(0, 1, 0, 1, 0),
        time=0:4,
        time.out=2.25,
        conc.blq="drop",
        interp.method=n
      ),
      interpolations[[n]],
      info=n
    )
  }
  # Ensure that the "log down" part works, too.
  expect_equal(
    interpolate.conc(
      conc=c(0, 1, 0.5, 1, 0),
      time=0:4,
      time.out=1.5,
      interp.method="lin up/log down"),
    exp(mean(log(c(1, 0.5))))
  )
  
  # Confirm that extrapolating before the first time uses conc.origin
  expect_equal(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=-1),
    0,
    info="conc.origin defaults to zero"
  )
  expect_equal(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=-1,
      conc.origin=NA),
    NA,
    info="conc.origin is honored as NA"
  )
  expect_equal(
    interpolate.conc(
      conc=0:1,
      time=0:1,
      time.out=-1,
      conc.origin=5
    ),
    5,
    info="conc.origin is honored as a number"
  )
  
  # expect_equal(
  #   interpolate.conc(
  #     conc=c(NA, 1, 2),
  #     time=c(0, 1, 2),
  #     time.out=1.5,
  #     check=FALSE
  #   ),
  #   1.5,
  #   info="Skipping the checks with an NA, but not bounding the interpolation gives the expected value."
  # )
  
  # by default, zeros in the middle are omitted
  expect_equal(
    interpolate.conc(
      conc=c(0, 1, 2, 0, 1, 0),
      time=0:5,
      time.out=3,
      conc.blq=NULL
    ),
    exp(mean(log(c(2, 1))))
  )
  # Zeros in the middle are accurately interpolated
  expect_equal(
    interpolate.conc(
      conc=c(0, 1, 2, 0, 1, 0),
      time=0:5,
      time.out=2.5,
      conc.blq=0
    ),
    1
  )
})

# extrapolate.conc ####

test_that("extrapolate.conc", {
  # Ensure that if Clast is NA that the extrapolated concentration is NA.
  expect_warning(
    v1 <-
      extrapolate.conc(
        conc=NA,
        time=1,
        time.out=2,
        extrap.method="AUCinf"
      ),
    regexp="All concentration data are missing"
  )
  expect_equal(v1, NA)

  # Extrapolating AUClast is always 0
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 0),
      time=1:3,
      time.out=4,
      extrap.method="AUClast"
      ),
    0
  )
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 0),
      time=1:3,
      time.out=2.5,
      extrap.method="AUClast"
    ),
    0
  )
  
  # Extrapolating AUCall after the last value and it is 0 is 0
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 0),
      time=1:3,
      time.out=4,
      extrap.method="AUCall"
    ),
    0
  )

  # Extrapolating AUCall after the last value and it is nonzero is 0
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 1),
      time=1:3,
      time.out=4,
      lambda.z=2,
      extrap.method="AUCall"
    ),
    0
  )

  # Extrapolating AUCall between Tlast and the first BLQ is a linear
  # interpolation.
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 0),
      time=1:3,
      time.out=2.5,
      extrap.method="AUCall"
    ),
    0.5
  )

  # Extrapolating AUCinf between the Clast value and a 0 ignores the
  # 0.
  expect_equal(
    extrapolate.conc(
      conc=c(0, 1, 0),
      time=1:3,
      time.out=2.5,
      lambda.z=1,
      extrap.method="AUCinf"
    ),
    1*exp(-1*0.5)
  )

  # A slightly less trivial example
  expect_equal(
    extrapolate.conc(
      conc=c(0, 5, 0),
      time=1:3,
      time.out=2.5,
      lambda.z=3,
      extrap.method="AUCinf"
    ),
    5*exp(-3*0.5)
  )

  # Extrapolating AUCinf with lambda.z=NA is NA.
  expect_equal(
    extrapolate.conc(
      conc=c(0, 5, 0),
      time=1:3,
      time.out=2.5,
      lambda.z=NA,
      extrap.method="AUCinf"
    ),
    as.numeric(NA)
  )

  # Extrapolating with all NA is NA.
  expect_warning(
    expect_equal(
      extrapolate.conc(
        conc=rep(NA, 3),
        time=1:3,
        time.out=2.5,
        lambda.z=NA,
        extrap.method="AUCinf"
      ),
      NA
    ),
    class = "pknca_conc_all_missing"
  )

  # Ensure that extrapolation beyond the last point works if the last point is 0
  extrapolations <-
    list(
      AUCinf=exp(-2),
      AUCall=0,
      AUClast=0
    )
  for (n in names(extrapolations)) {
    expect_equal(
      extrapolate.conc(
        conc=c(0, 1, 0.5, 1, 0),
        time=0:4, 5,
        lambda.z=1,
        extrap.method=n
      ),
      extrapolations[[n]],
      info=n
    )
  }

  # Ensure that extrapolation to the last point of 0 works
  extrapolations <-
    list(
      AUCinf=exp(-1),
      AUCall=0,
      AUClast=0
    )
  for (n in names(extrapolations)) {
    expect_equal(
      extrapolate.conc(
        conc=c(0, 1, 0.5, 1, 0),
        time=0:4, 4,
        lambda.z=1,
        extrap.method=n
      ),
      extrapolations[[n]],
      info=n
    )
  }

  # Ensure that extrapolation between the last points with the last point 0 works
  extrapolations <-
    list(
      AUCinf=exp(-0.5),
      AUCall=0.5,
      AUClast=0
    )
  for (n in names(extrapolations)) {
    expect_equal(
      extrapolate.conc(
        conc=c(0, 1, 0.5, 1, 0),
        time=0:4, 3.5,
        lambda.z=1,
        extrap.method=n
      ),
      extrapolations[[n]],
      info=n
    )
  }
})

# interp.extrap.conc ####

test_that("interp.extrap.conc", {
  # All zeros always interpolates to all zeros
  expect_equal(
    interp.extrap.conc(
      conc=c(0, 0, 0),
      time=0:2,
      time.out=c(0, 0.5, 2.5),
      interp.method="lin up/log down"
    ),
    rep(0, 3)
  )
  
  # Ensure that data checking works correctly
  expect_equal(
    interp.extrap.conc(
      conc=c(0, 1, 0.5, 1, 0),
      time=0:4,
      time.out=1.5,
      interp.method="lin up/log down"
    ),
    exp(mean(log(c(1, 0.5))))
  )

  # Ensure that NA for time.out results in NA output
  expect_warning(
    expect_equal(
      interp.extrap.conc(
        conc=c(0, 1, 0.5, 1, 0),
        time=0:4,
        time.out=c(1.5, NA),
        interp.method="lin up/log down"
      ),
      c(exp(mean(log(c(1, 0.5)))), NA)
    )
  )

  # Ensure a warning with NA for time.out
  expect_warning(
    interp.extrap.conc(
      conc=c(0, 1, 0.5, 1, 0),
      time=0:4,
      time.out=c(1.5, NA),
      interp.method="lin up/log down"
    ),
    regexp="An interpolation/extrapolation time is NA"
  )
})

# interp.extrap.conc.dose ####

test_that("interp.extrap.conc.dose handles all eventualities", {
  event_choices <- unlist(event_choices_interp.extrap.conc.dose, use.names=FALSE)
  eventualities <-
    expand.grid(
      event_before=setdiff(event_choices, "output_only"),
      event=setdiff(event_choices, "none"),
      event_after=setdiff(event_choices, "output_only")
    )
  eventualities$method <- NA_character_
  for (nm in names(interp.extrap.conc.dose.select)) {
    mask_selected <-
      do.call(
        interp.extrap.conc.dose.select[[nm]]$select,
        list(x=eventualities)
      )
    expect_true(
      any(mask_selected),
      info=sprintf("interp.extrap.conc.dose.select[[%s]] matched at least one eventuality", nm)
    )
    expect_true(
      !any(mask_selected & !is.na(eventualities$method)),
      info=sprintf("interp.extrap.conc.dose.select[[%s]] overlapped with another method.", nm)
    )
    eventualities$method[mask_selected] <- nm
    # if (interactive()) {
    #   cat(sprintf("%d/%d rows filled at %s\n", sum(!is.na(eventualities$method)), nrow(eventualities), nm))
    # }
  }
  expect_false(
    any(is.na(eventualities$method)),
    info="interp.extrap.conc.dose.select matched all eventualities"
  )
})

test_that("interp.extrap.conc.dose", {
  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=-2,
      check=FALSE
    ),
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=-2
    ),
    info="Check is respected"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=-2
    ),
    structure(0, Method="Before all events"),
    info="Interpolation before all events yields conc.origin which defaults to zero."
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      conc.origin=NA,
      time.out=-2),
    structure(NA_real_, Method="Before all events"),
    info="Interpolation before all events yields conc.origin respecting its input."
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=-1
    ),
    structure(0, Method="Observed concentration"),
    info="When there is a concentration measurement at a time point, it is returned."
  )
  
  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=-0.1
    ),
    structure(0, Method="Extrapolation"),
    info="When the previous measurement is zero and there is no dose between, it is returned."
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=0),
    structure(0, Method="Extrapolation"),
    info="When the previous measurement is zero it is at the time of the dose, zero is returned."
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=0.1
    ),
    structure(0.1, Method="Dose before, concentration after without a dose"),
    info="Extrapolation to a dose then interpolation between the dose and the next time works."
  )

  # expect_equal(
  #   expect_warning(
  #       interp.extrap.conc.dose(
  #         conc=c(0, 1, 2, 1, 0.5, 0.25),
  #         time=c(-1, 1:5),
  #         time.dose=c(0, 0.1),
  #         time.out=0.2
  #       ),
  #       regexp="Cannot interpolate between two doses or after a dose without a concentration after the first dose.",
  #       fixed=TRUE,
  #       info="Two doses in a row generates a warning"
  #   ),
  #   structure(NA_real_, Method="Dose before, concentration after without a dose"),
  #   info="Extrapolation to a dose then interpolation between the dose and the next time gives NA when the dose is NA."
  # )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=5
    ),
    structure(0.25, Method="Observed concentration"),
    info="Copy from after the dose."
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=7
    ),
    structure(NA_real_, Method="Extrapolation"),
    info="Extrapolation without lambda.z gives NA result"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=7,
      lambda.z=log(2)
    ),
    structure(0.0625, Method="Extrapolation"),
    info="Extrapolation with lambda.z gives result"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=0:2,
      time=0:2,
      time.dose=0,
      time.out=0.5
    ),
    structure(0.5, Method="Interpolation"),
    info="Interpolation works"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0:2, 1),
      time=0:3,
      time.dose=0,
      time.out=2.5,
      method="linear"
    ),
    structure(sqrt(2), Method="Interpolation"),
    info="Interpolation respects method"
  )
  
  expect_equal(
    interp.extrap.conc.dose(
      conc=0:2,
      time=0:2,
      time.dose=0,
      route.dose="intravascular",
      time.out=0,
      duration.dose=0,
      out.after=FALSE
    ),
    structure(0, Method="Observed concentration"),
    info="Observed before IV bolus"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=0:2,
      time=0:2,
      time.dose=0,
      route.dose="intravascular",
      time.out=0,
      duration.dose=0,
      out.after=TRUE
    ),
    structure(1, Method="Immediately after an IV bolus with a concentration next"),
    info="Observed after IV bolus, one concentration"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 2, 1),
      time=0:2,
      time.dose=0,
      route.dose="intravascular",
      time.out=0,
      duration.dose=0,
      out.after=TRUE
    ),
    structure(4, Method="Immediately after an IV bolus with a concentration next"),
    info="Observed after IV bolus, two concentrations"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=c(2, 1),
      time=1:2,
      time.dose=0,
      route.dose="intravascular",
      time.out=0.5,
      duration.dose=0
    ),
    structure(2*sqrt(2), Method="After an IV bolus with a concentration next"),
    info="After IV bolus, two concentrations"
  )

  expect_equal(
    interp.extrap.conc.dose(
      conc=2,
      time=1,
      time.dose=0,
      route.dose="intravascular",
      time.out=0.5,
      duration.dose=0
    ),
    structure(2, Method="After an IV bolus with a concentration next"),
    info="After IV bolus, one concentration"
  )
  
  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=c(-2, 2)
      ),
    structure(
      c(0, 2),
      Method=c("Before all events", "Observed concentration")
    ),
    info="Outputs are in the same order as inputs (initially sorted)"
  )
  
  expect_equal(
    interp.extrap.conc.dose(
      conc=c(0, 1, 2, 1, 0.5, 0.25),
      time=c(-1, 1:5),
      time.dose=0,
      time.out=c(2, -2)
    ),
    structure(
      c(2, 0),
      Method=c("Observed concentration", "Before all events")
    ),
    info="Outputs are in the same order as inputs (reverse sorted time.out)"
  )
})