Skip to contents

Morphine (Wang 2013)

Source: vignettes/articles/Wang_2013_morphine.Rmd
Wang_2013_morphine.Rmd

Model and source

  • Citation: Wang C, Sadhasivam S, Krekels EHJ, Dahan A, Tibboel D, Danhof M, Vinks AA, Knibbe CAJ (2013). Developmental changes in morphine clearance across the entire paediatric age range are best described by a bodyweight-dependent exponent model. Clinical Drug Investigation 33(7):523-534. doi:10.1007/s40261-013-0097-6. PMID:23754691. DDMORE Foundation Model Repository: DDMODEL00000269 (Model I, morphine alone).
  • Description: Two-compartment population PK model for morphine across the entire paediatric age range and adults using a bodyweight-dependent allometric exponent (BDE) on clearance, with adolescent-specific intercompartmental clearance and central volume and an adult-stratum oral-bioavailability adjustment, as packaged in DDMORE Foundation Model Repository entry DDMODEL00000269 (Wang 2013 Model I).
  • Article: Clinical Drug Investigation 2013;33(7):523-534
  • DDMORE Foundation Model Repository entry: DDMODEL00000269 – Model I (morphine alone). The same bundle ships a separate joint morphine + M3G model (Model II) which is not packaged here; see Assumptions and deviations.

Population

The publication abstract reports a pooled paediatric and adult population of “358 neonates, infants, children and adults” plus “117 adolescents” (475 subjects total) used to develop a population PK model for morphine. The DDMORE bundle’s .mod $INPUT comments label each subject with a POP value that codes the age stratum:

  • POP = 1 – newborns / infants / young children, 0-3 years (encoded here as CHILD = 1).
  • POP = 2 – children-adolescents, 6-15 years (encoded here as ADOLESCENT = 1).
  • POP = 3 – adults, 18-36 years (encoded here as CHILD = 0 AND ADOLESCENT = 0, the implicit baseline).

Body weights in the bundle’s simulated dataset span 0.6 kg (term neonate) to 85 kg (adult). The original Wang 2013 PDF was not on disk under mab_human_consensus/literature/ at extraction time, so per-study counts, sex / race breakdowns, and indication-specific information could not be cross-checked against the publication; everything in this section comes from the abstract (PMID 23754691) and the .mod $INPUT comments.

The same information is available programmatically via readModelDb("Wang_2013_morphine")$population.

Source trace

Per-parameter origin is recorded as an in-file comment next to each ini() entry in inst/modeldb/ddmore/Wang_2013_morphine.R. The table below collects them for review. All parameter values come from the FINAL PARAMETER ESTIMATE block of Output_real_ModelI_Morphine.lst (the real-data NONMEM listing reporting MINIMIZATION SUCCESSFUL). The structural equations come from the Executable_ModelI1_Morphine.mod $PK and $ERROR blocks.

Equation / parameter Value Source location
lkdec log(0.594) (KDEC, unitless) Output_real_ModelI_Morphine.lst THETA TH 1
lkmin log(0.872) (KMAX - KDEC, unitless) Output_real_ModelI_Morphine.lst THETA TH 2
lkhal log(4.01) (KHAL, kg) Output_real_ModelI_Morphine.lst THETA TH 3
lgamma log(4.62) (GAMMA, unitless) Output_real_ModelI_Morphine.lst THETA TH 4
lcl log(1.62) L/min @ 70 kg adult Output_real_ModelI_Morphine.lst THETA TH 5
lq log(1.90) L/min @ 70 kg, non-adolescent Output_real_ModelI_Morphine.lst THETA TH 6
lvc log(81.2) L @ 70 kg, non-adolescent Output_real_ModelI_Morphine.lst THETA TH 7
lvp log(128) L @ 70 kg Output_real_ModelI_Morphine.lst THETA TH 8
propSd 0.432 (NONMEM log-transform-both-sides residual SD; equivalent to proportional in linear nlmixr2 space) Output_real_ModelI_Morphine.lst THETA TH 9
lq_adolescent log(0.500) L/min, NOT BW-scaled Output_real_ModelI_Morphine.lst THETA TH10
lvc_adolescent log(46.0) L @ 70 kg, BW-scaled Output_real_ModelI_Morphine.lst THETA TH11
e_age_adult_f 0.88 (adult-stratum F1) Executable_ModelI1_Morphine.mod $PK IF (POP.EQ.3) F1 = 0.88
etalcl IIV 0.159 (variance, log scale) Output_real_ModelI_Morphine.lst OMEGA ETA1 (CL)
etalvc IIV 0.253 (variance, log scale) Output_real_ModelI_Morphine.lst OMEGA ETA3 (V1)
Q / V2 IIV 0 (FIXED in $OMEGA) Output_real_ModelI_Morphine.lst OMEGA ETA2, ETA4
Structural equations ADVAN5 2-compartment with K10 = CL/V1, K12 = Q/V1, K21 = Q/V2 Executable_ModelI1_Morphine.mod $MODEL + $PK + $SUBROUTINE ADVAN5
BDE allometric form KBDE(WT) = KMAX - KDEC x WT^GAMMA / (KHAL^GAMMA + WT^GAMMA) with KMAX = KDEC + (KMAX - KDEC) = 0.594 + 0.872 = 1.466 Wang 2013 abstract (“decreases sigmoidal across paediatric range from 1.47 to 0.88, midpoint 4.01 kg”) and Executable_ModelI1_Morphine.mod $PK
Concentration units ug/L .mod $INPUT comment CONC ;ug/L
Time units minutes .mod $INPUT comment TIME ;in min
Dose units micrograms (ug) .mod $INPUT comment AMT ;in microgram (ug)

BDE relationship between bodyweight and CL exponent

The publication’s headline result is that the allometric exponent on clearance changes sigmoidally with body weight, dropping from ~1.47 (neonates) to 0.88 (adults), with the half-maximal effect at 4.01 kg. Plot KBDE vs body weight directly from the model parameters to confirm the shape:

ini_vals <- list(KDEC = 0.594, KMIN = 0.872, KHAL = 4.01, GAMMA = 4.62)
wt_grid <- 10 ^ seq(log10(0.5), log10(120), length.out = 200)
kbde <- with(ini_vals,
  (KDEC + KMIN) - KDEC * wt_grid ^ GAMMA / (KHAL ^ GAMMA + wt_grid ^ GAMMA))
ggplot(data.frame(WT = wt_grid, kbde = kbde), aes(WT, kbde)) +
  geom_line(linewidth = 0.8) +
  geom_hline(yintercept = c(0.872, 1.466), linetype = "dotted", colour = "grey40") +
  geom_vline(xintercept = 4.01, linetype = "dotted", colour = "grey40") +
  annotate("text", x = 0.55, y = 1.466, label = "KMAX = 1.466",
    hjust = 0, vjust = -0.4, size = 3) +
  annotate("text", x = 0.55, y = 0.872, label = "KMIN = 0.872",
    hjust = 0, vjust = -0.4, size = 3) +
  annotate("text", x = 4.01, y = 1.20, label = "KHAL = 4.01 kg",
    hjust = -0.05, size = 3) +
  scale_x_log10() +
  labs(x = "Body weight (kg, log scale)", y = "BDE allometric exponent on CL",
    title = "Bodyweight-dependent exponent (BDE) on morphine CL",
    caption = "Reproduces the Wang 2013 abstract finding (exponent drops from ~1.47 to 0.88, half-max at 4.01 kg).")

Virtual cohort

Build a deterministic typical-subject cohort and a stochastic cohort covering the three age strata. WT distributions per stratum mirror the weight bins observed in the bundle’s simulated dataset (0.6-9 kg for infants / young children, 20-70 kg for adolescents, 60-85 kg for adults).

set.seed(20130523) # publication month
n_per_stratum <- 100L

make_subjects <- function(n, wt_min, wt_max, child, adol, label, id_offset) {
  tibble::tibble(
    id         = id_offset + seq_len(n),
    WT         = exp(runif(n, log(wt_min), log(wt_max))),
    CHILD      = as.integer(child),
    ADOLESCENT = as.integer(adol),
    treatment  = factor(label,
      levels = c("Infant / young child (POP=1)",
                 "Adolescent (POP=2)",
                 "Adult (POP=3)"))
  )
}

cohort <- dplyr::bind_rows(
  make_subjects(n_per_stratum, wt_min = 0.6, wt_max = 9.0,
    child = 1L, adol = 0L,
    label = "Infant / young child (POP=1)", id_offset = 0L),
  make_subjects(n_per_stratum, wt_min = 20,  wt_max = 70,
    child = 0L, adol = 1L,
    label = "Adolescent (POP=2)",            id_offset = n_per_stratum),
  make_subjects(n_per_stratum, wt_min = 60,  wt_max = 90,
    child = 0L, adol = 0L,
    label = "Adult (POP=3)",                 id_offset = 2L * n_per_stratum)
)

A single 60-min IV-equivalent infusion of weight-scaled morphine (100 ug/kg total dose) is administered to each subject and concentrations are sampled densely over 8 h (480 min). The dose level is illustrative – Wang 2013 pooled across many study-specific regimens; the goal here is to demonstrate the BDE-driven CL scaling and the adolescent / adult parameter overrides.

sim_dur <- 60          # infusion duration (min)
sim_end <- 480         # observation horizon (min)
dose_per_kg <- 100     # ug/kg

dose_rows <- cohort |>
  dplyr::mutate(
    time = 0,
    amt  = dose_per_kg * WT,
    rate = (dose_per_kg * WT) / sim_dur,
    cmt  = "central",
    evid = 1L
  )

obs_times <- c(seq(0, 60, by = 5), seq(75, sim_end, by = 15))
obs_rows <- cohort |>
  tidyr::crossing(time = obs_times) |>
  dplyr::mutate(amt = 0, rate = 0, cmt = NA_character_, evid = 0L)

events <- dplyr::bind_rows(dose_rows, obs_rows) |>
  dplyr::select(id, time, amt, rate, cmt, evid,
                WT, CHILD, ADOLESCENT, treatment) |>
  dplyr::arrange(id, time, dplyr::desc(evid))

stopifnot(!anyDuplicated(unique(events[, c("id", "time", "evid")])))

Simulation 

mod <- rxode2::rxode2(readModelDb("Wang_2013_morphine"))
#> ℹ parameter labels from comments will be replaced by 'label()'
conc_unit <- mod$units[["concentration"]]
sim <- rxode2::rxSolve(
  mod, events = events,
  keep = c("WT", "CHILD", "ADOLESCENT", "treatment")
)

Replicate published findings

Typical-value concentration profiles by stratum

The deterministic typical-value (zeroRe()) profiles below show how the adolescent overrides on Q and V1 and the adult F1 = 0.88 reduce concentrations relative to a strict allometric model.

typical_subjects <- tibble::tibble(
  id         = 1:6,
  WT         = c(1, 5, 30, 50, 70, 85),
  CHILD      = c(1L, 1L, 0L, 0L, 0L, 0L),
  ADOLESCENT = c(0L, 0L, 1L, 1L, 0L, 0L),
  treatment  = factor(
    c("0-3 yr, 1 kg", "0-3 yr, 5 kg",
      "6-15 yr, 30 kg", "6-15 yr, 50 kg",
      "Adult, 70 kg",  "Adult, 85 kg"),
    levels = c("0-3 yr, 1 kg", "0-3 yr, 5 kg",
               "6-15 yr, 30 kg", "6-15 yr, 50 kg",
               "Adult, 70 kg",  "Adult, 85 kg"))
)

typical_doses <- typical_subjects |>
  dplyr::mutate(
    time = 0,
    amt  = dose_per_kg * WT,
    rate = (dose_per_kg * WT) / sim_dur,
    cmt  = "central",
    evid = 1L
  )

typical_obs <- typical_subjects |>
  tidyr::crossing(time = seq(0, sim_end, by = 2)) |>
  dplyr::mutate(amt = 0, rate = 0, cmt = NA_character_, evid = 0L)

typical_events <- dplyr::bind_rows(typical_doses, typical_obs) |>
  dplyr::select(id, time, amt, rate, cmt, evid,
                WT, CHILD, ADOLESCENT, treatment) |>
  dplyr::arrange(id, time, dplyr::desc(evid))

mod_typical <- mod |> rxode2::zeroRe()
sim_typical <- rxode2::rxSolve(
  mod_typical, events = typical_events,
  keep = c("WT", "CHILD", "ADOLESCENT", "treatment")
)
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalvc'
#> Warning: multi-subject simulation without without 'omega'

sim_typical |>
  dplyr::filter(!is.na(Cc), time > 0) |>
  ggplot(aes(time, Cc, colour = treatment)) +
  geom_line(linewidth = 0.7) +
  scale_y_log10() +
  labs(x = "Time post-dose (min)",
       y = paste0("Morphine concentration (", conc_unit, ", log scale)"),
       colour = NULL,
       title = "Typical-value morphine concentration after a 100 ug/kg, 60-min infusion",
       caption = "Across the three age strata; deterministic prediction (zeroRe).") +
  theme_minimal()

VPC-style cohort summary by stratum 

sim |>
  dplyr::filter(!is.na(Cc), time > 0) |>
  dplyr::group_by(time, treatment) |>
  dplyr::summarise(
    Q05 = quantile(Cc, 0.05, na.rm = TRUE),
    Q50 = quantile(Cc, 0.50, na.rm = TRUE),
    Q95 = quantile(Cc, 0.95, na.rm = TRUE),
    .groups = "drop"
  ) |>
  ggplot(aes(time, Q50)) +
  geom_ribbon(aes(ymin = Q05, ymax = Q95), alpha = 0.25) +
  geom_line() +
  facet_wrap(~ treatment, scales = "free_y") +
  labs(x = "Time post-dose (min)",
       y = paste0("Cc (", conc_unit, ")"),
       title = "5-50-95 percentiles by age stratum",
       caption = paste("Each panel: 100 simulated subjects;",
                       "100 ug/kg over 60 min."))

PKNCA validation

Compute Cmax / Tmax / AUClast / AUCinf / half-life over the 8-h post-infusion window. The treatment grouping (treatment) is placed before id in the formula per the project convention so per-stratum NCA results roll up automatically.

sim_nca <- sim |>
  dplyr::filter(!is.na(Cc)) |>
  dplyr::select(id, time, Cc, treatment)

dose_df <- events |>
  dplyr::filter(evid == 1) |>
  dplyr::select(id, time, amt, treatment)

conc_obj <- PKNCA::PKNCAconc(sim_nca, Cc ~ time | treatment + id)
dose_obj <- PKNCA::PKNCAdose(dose_df, amt ~ time | treatment + id)

intervals <- data.frame(
  start      = 0,
  end        = sim_end,
  cmax       = TRUE,
  tmax       = TRUE,
  auclast    = TRUE,
  aucinf.obs = TRUE,
  half.life  = TRUE
)

nca_data <- PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals)
nca_res  <- PKNCA::pk.nca(nca_data)
#>  ■■■■■■■■■                         27% |  ETA:  7s
#>  ■■■■■■■■■■■■■■■■■■■               59% |  ETA:  4s
#>  ■■■■■■■■■■■■■■■■■■■■■■■■■■■■      90% |  ETA:  1s
knitr::kable(summary(nca_res),
  caption = "Simulated NCA by age stratum (100 ug/kg, 60-min infusion).")
Simulated NCA by age stratum (100 ug/kg, 60-min infusion).
start end treatment N auclast cmax tmax half.life aucinf.obs
0 480 Infant / young child (POP=1) 100 7070 [58.6] 40.8 [31.3] 60.0 [60.0, 60.0] 384 [297] 11300 [111]
0 480 Adolescent (POP=2) 100 3810 [38.6] 42.7 [30.6] 60.0 [60.0, 60.0] 164 [49.3] 4080 [43.3]
0 480 Adult (POP=3) 100 3480 [32.3] 29.6 [27.3] 50.0 [50.0, 55.0] 136 [46.1] 3790 [38.1]

The Wang 2013 publication does not tabulate per-subject or per-stratum NCA values for direct numerical comparison (the publication focuses on clearance scaling, not exposure metrics), so the side-by-side “comparison-against-published-NCA” sanity check normally rendered here is deferred to the F.2 self-consistency check below.

F.2 self-consistency check vs the bundled simulated dataset

Simulated_DataModel1_Morphine.csv ships with the DDMORE bundle. The bundle’s Output_simulated_ModelI_Morphine.lst re-fits the model on this dataset; the recorded CONC column is what the original investigators derived by forward-simulating with the final estimates. Re-simulate the same event records with the typical-value model and confirm the trajectories track the recorded CONC values.

bundle_csv <- system.file("modeldb", package = "nlmixr2lib") # not used; explicit path below
sim_csv <- file.path(
  "/home/bill/github/mab_human_consensus/literature/from_people/ddmore",
  "ddmore_scraping/269/Simulated_DataModel1_Morphine.csv"
)
self_consistency_available <- file.exists(sim_csv)
# `.` is the missingness sentinel in the bundle's CSV; coerce numeric columns
# explicitly so we don't pick up character types via read.csv's heuristic.
raw <- utils::read.csv(sim_csv, na.strings = c(".", "NA", ""))
num_cols <- intersect(c("AMT", "RATE", "DV", "CONC", "BW", "TIME"), names(raw))
raw[num_cols] <- lapply(raw[num_cols], function(x) suppressWarnings(as.numeric(x)))

bundle <- raw |>
  dplyr::transmute(
    id   = as.integer(ID),
    time = as.numeric(TIME),
    amt  = as.numeric(ifelse(MDV == 1, AMT, 0)),
    rate = as.numeric(ifelse(MDV == 1, RATE, 0)),
    cmt  = ifelse(MDV == 1, "central", NA_character_),
    evid = ifelse(MDV == 1, 1L, 0L),
    WT   = as.numeric(BW),
    CHILD      = as.integer(POP == 1),
    ADOLESCENT = as.integer(POP == 2),
    obs_DV   = ifelse(MDV == 0, DV, NA_real_),
    obs_CONC = ifelse(MDV == 0, CONC, NA_real_),
    POP        = POP
  )

# Add a denser observation grid alongside the bundle's sparse points so the
# typical-value trajectory is visible in the plot.
bundle_dense_obs <- bundle |>
  dplyr::distinct(id, WT, CHILD, ADOLESCENT, POP) |>
  tidyr::crossing(time = c(seq(0, 360, by = 5), seq(420, 5400, by = 60))) |>
  dplyr::mutate(amt = 0, rate = 0, cmt = NA_character_, evid = 0L,
                obs_DV = NA_real_, obs_CONC = NA_real_)

bundle_events <- dplyr::bind_rows(bundle, bundle_dense_obs) |>
  dplyr::arrange(id, time, dplyr::desc(evid))

# Drop columns rxode2 doesn't accept on event records before passing to rxSolve.
sim_bundle <- rxode2::rxSolve(
  mod_typical,
  events = bundle_events |>
    dplyr::select(id, time, amt, rate, cmt, evid, WT, CHILD, ADOLESCENT),
  keep = c("WT", "CHILD", "ADOLESCENT")
) |> as.data.frame()

# Pair the bundle's recorded CONC observations against the simulated
# trajectory at matching (id, time).
bundle_obs_only <- bundle |>
  dplyr::filter(!is.na(obs_CONC)) |>
  dplyr::select(id, time, WT, POP, obs_CONC)

stratum_label <- function(POP) {
  factor(dplyr::case_when(
    POP == 1 ~ "Infant / young child (POP=1)",
    POP == 2 ~ "Adolescent (POP=2)",
    POP == 3 ~ "Adult (POP=3)"
  ),
  levels = c("Infant / young child (POP=1)",
             "Adolescent (POP=2)",
             "Adult (POP=3)"))
}

bundle_obs_only$stratum <- stratum_label(bundle_obs_only$POP)
sim_bundle$stratum <- stratum_label(
  ifelse(sim_bundle$CHILD == 1, 1L,
  ifelse(sim_bundle$ADOLESCENT == 1, 2L, 3L)))

ggplot() +
  geom_line(data = dplyr::filter(sim_bundle, time > 0, !is.na(Cc)),
    aes(time, Cc, group = id, colour = stratum), alpha = 0.5) +
  geom_point(data = bundle_obs_only,
    aes(time, obs_CONC), shape = 21, fill = "white", colour = "black", size = 2) +
  scale_y_log10() +
  facet_wrap(~ stratum, scales = "free") +
  labs(x = "Time post-first-dose (min)",
       y = paste0("Cc (", conc_unit, "), log scale"),
       colour = NULL,
       title = "F.2 self-consistency: typical-value re-simulation vs bundle CONC",
       caption = paste("Lines: this model's typical-value (zeroRe) prediction;",
                       "Points: bundle Simulated_DataModel1_Morphine.csv CONC column."))
# Summarise the per-observation pred-vs-obs offset (typical-value, no IIV).
match_tbl <- bundle_obs_only |>
  dplyr::inner_join(
    sim_bundle |>
      dplyr::filter(!is.na(Cc)) |>
      dplyr::select(id, time, Cc),
    by = c("id", "time")
  ) |>
  dplyr::mutate(rel_err = (Cc - obs_CONC) / obs_CONC)

knitr::kable(
  match_tbl |>
    dplyr::group_by(stratum) |>
    dplyr::summarise(
      n       = dplyr::n(),
      median_rel_err = sprintf("%+.2f", median(rel_err, na.rm = TRUE)),
      q05_q95 = sprintf("%+.2f / %+.2f",
        quantile(rel_err, 0.05, na.rm = TRUE),
        quantile(rel_err, 0.95, na.rm = TRUE))
    ) |>
    dplyr::rename(
      Stratum                 = stratum,
      `N matched obs`         = n,
      `Median (Cc - obs)/obs` = median_rel_err,
      `5th / 95th pct`        = q05_q95
    ),
  caption = paste("F.2 self-consistency: relative error of the typical-value",
    "re-simulation against the bundle's recorded CONC values."))

The bundle’s CONC values include between-subject variability (they were generated with the full IIV / residual-error model), while the re-simulation here uses typical-value parameters with no IIV and no residual error. A typical-value point can therefore differ from a per-subject recorded CONC by tens of percent (especially for the small-N subjects in the simulated bundle), which is expected variability rather than a miscoded equation.

Assumptions and deviations

  • Model I only. The DDMORE bundle ships two executables for this publication: Model I (morphine alone, packaged here) and Model II (joint morphine + M3G metabolite PK with separate clearance pathways). The task assigned a singular target filename Wang_2013_morphine.R; an operator follow-up confirmed Model I is in scope. Anyone needing the parent + M3G joint model should look at ddmore_scraping/269/Executable_ModelII_MM3G.mod directly – it is structurally a 3-compartment model (morphine 2-comp + M3G 1-comp) with metabolic-pathway clearances and would warrant a separate Wang_2013b_morphine.R extraction.
  • Adult oral-bioavailability arm. The .mod $PK block applies F1 = 0.88 only when POP = 3 (adults). The publication abstract does not describe the route of administration in detail, and the original Wang 2013 PDF was not on disk under mab_human_consensus/literature/ at extraction time, so the rationale for the 0.88 reduction in adults could not be cross-checked against the paper text. The model here preserves the .mod’s F1 logic verbatim (f(central) = 0.88 whenever CHILD = 0 AND ADOLESCENT = 0) and exposes the value as e_age_adult_f so it is overridable.
  • POP encoding. The source data column POP (1 = 0-3 yr, 2 = 6-15 yr, 3 = 18-36 yr) is decomposed into the canonical binary indicators CHILD (POP = 1) and ADOLESCENT (POP = 2). Adults are the implicit baseline (CHILD = 0 AND ADOLESCENT = 0). This matches the encoding used by CarlssonPetri_2021_liraglutide. The decomposition is defined in the model file’s covariateData$CHILD$notes and covariateData$ADOLESCENT$notes.
  • NONMEM log-transform-both-sides residual error. The .mod $ERROR block sets Y = LOG(F) + ERR(1) * W with $SIGMA EPS1 FIXED = 1 and W = THETA(9) = 0.432. NONMEM “additive on log-scale with SIGMA fixed at 1” maps to proportional residual error in nlmixr2’s linear space with propSd = 0.432 (see naming-conventions.md Section “NONMEM -> nlmixr2 syntax translation”).
  • Real vs simulated executable. The Output_real_*.lst reports an additional EPS(2) term gated by IF (TIME.GT.1900.AND.FLAG.EQ.{1,2}) to handle a study-specific assay artefact after a 1900-min cutoff (variance 0.455). The Executable_ModelI1_Morphine.mod (the simulation- ready executable shipped in the bundle) drops this term and uses only Y = LOG(F) + ERR(1) * W. We follow the simulation-ready executable here; the FLAG-conditional error is a property of one of the original studies’ assays and is not portable to a generic library model.
  • Q and V2 IIV. The source $OMEGA fixes IIV on Q and V2 to zero (0 FIX). Only CL and V1 carry IIV in this model.
  • Simulated dataset is intentionally minimal. The bundle’s Simulated_DataModel1_Morphine.csv has only 14 subjects (8 in POP=1, 4 in POP=2, 2 in POP=3); it is a regression-style smoke test, not a representative population. The “Virtual cohort” section above builds a larger cohort spanning the publication’s reported BW range for the VPC and PKNCA panels.
  • External cross-check. No publication PDF is on disk under mab_human_consensus/literature/ at extraction time, so per-subject parameter estimates and tabulated NCA values from the publication could not be compared against the simulation. The F.2 self-consistency check above is the substitute mandated by references/ddmore-source.md Section “Validation strategy by model type” for DDMORE-source models without an on-disk publication.