| Title: | Create Visual Predictive Checks |
| Version: | 1.2.2 |
| Date: | 2021-01-11 |
| Author: | Ron Keizer <ronkeizer@gmail.com> |
| Maintainer: | Ron Keizer <ronkeizer@gmail.com> |
| Description: | Visual predictive checks are a commonly used diagnostic plot in pharmacometrics, showing how certain statistics (percentiles) for observed data compare to those same statistics for data simulated from a model. The package can generate VPCs for continuous, categorical, censored, and (repeated) time-to-event data. |
| Depends: | R (≥ 3.1.0) |
| Imports: | classInt, dplyr, MASS, survival, ggplot2, readr, stringr, tidyr |
| License: | MIT + file LICENSE |
| LazyData: | true |
| URL: | https://github.com/ronkeizer/vpc |
| Suggests: | knitr, testit |
| RoxygenNote: | 7.1.0 |
| NeedsCompilation: | no |
| Packaged: | 2021-01-11 19:53:47 UTC; ronkeizer |
| Repository: | CRAN |
| Date/Publication: | 2021-01-11 20:20:02 UTC |
VPC package
Description
Create Visual Predictive Checks in R
Author(s)
Ron Keizer ronkeizer@gmail.com
Add noise / residual error to data
Description
Add noise / residual error to data
Usage
add_noise(x, ruv = list(proportional = 0, additive = 0, exponential = 0))
Arguments
x |
data |
ruv |
list describing the magnitude of errors. List arguments: "proportional", "additive", "exponential". |
Examples
library(dplyr)
ipred <- c(10, 8, 6, 4, 2, 0) %>% add_noise(ruv = list(proportional = 0.1, additive = 0.2))
Add sim index number
Description
Add simulation index number to simulation when not present
Usage
add_sim_index_number(sim, id = "id", sim_label = "sim")
Arguments
sim |
a data.frame containing the simulation data |
id |
character specifying the column name in the data.frame |
sim_label |
label to indicate simulation index (if available) |
Adds stratification to data set
Description
Adds stratification to data set
Usage
add_stratification(dat, strat, verbose = FALSE)
Arguments
dat |
data.frame |
strat |
vector of stratification variables |
verbose |
verbosity ('TRUE' or 'FALSE') |
Calculate appropriate bin separators for vpc
Description
This function calculates bin separators either using R's native binning approaches available in the classInt library such as 'kmeans', 'jenks', 'pretty' etc. Alternatively, a custom approach is available which is based on finding the nadirs in the density functions for the independent variable. Default approach is k-means clustering.
Usage
auto_bin(dat, type = "kmeans", n_bins = 8, verbose = FALSE, ...)
## S3 method for class 'numeric'
auto_bin(dat, type = "kmeans", n_bins = 8, verbose = FALSE, ...)
## S3 method for class 'data.frame'
auto_bin(dat, type = "kmeans", n_bins = 8, verbose = FALSE, ...)
Arguments
dat |
data frame |
type |
auto-binning type: "density", "time", or "data" |
n_bins |
number of bins to use; either a positive integer or "auto". For "density" the function might not return a solution with the exact number of bins. |
verbose |
show warnings and other messages (TRUE or FALSE) |
... |
arguments passed on to underlying binning functions |
Value
A vector of bin separators
Function to bin data based on a vector of bin separators, e.g. for use in VPC
Description
Function to bin data based on a vector of bin separators, e.g. for use in VPC
Usage
bin_data(x, bins = c(0, 3, 5, 7), idv = "time", labeled = F)
Arguments
x |
data |
bins |
numeric vector specifying bin separators |
idv |
variable in the data specifies the independent variable (e.g. "time") |
labeled |
whether a labeled factor instead of integers should be returned |
Check whether stratification columns are available
Description
Check whether stratification columns are available
Usage
check_stratification_columns_available(data, stratify, type = "observation")
Arguments
data |
'data.frame' with observation or simulation data |
stratify |
vector of stratification columns |
type |
either 'observation' or 'simulation' |
Compute Kaplan-Meier statistics
Description
Compute Kaplan-Meier statistics
Usage
compute_kaplan(
dat,
strat = "strat",
reverse_prob = FALSE,
rtte_conditional = TRUE,
ci = NULL
)
Arguments
dat |
data.frame with events |
strat |
vector of stratification variables |
reverse_prob |
reverse the probability (i.e. return '1-probability')? |
rtte_conditional |
'TRUE' (default) or 'FALSE'. Compute the probability for each event newly ('TRUE'), or calculate the absolute probability ('FALSE', i.e. the "probability of a 1st, 2nd, 3rd event etc" rather than the "probability of an event happening"). |
ci |
confidence interval to calculate, numeric vector of length 2 |
Compute KMMC statistics
Description
Kaplan-Meier Mean Covariate plots are a simulation-based diagnostic to study the influence of covariates and identify potential model misspecification.
Usage
compute_kmmc(dat, strat = NULL, reverse_prob = FALSE, kmmc = "DOSE")
Arguments
dat |
data.frame with events |
strat |
vector of stratification variables |
reverse_prob |
reverse the probability (i.e. return '1-probability')? |
kmmc |
variable to create the KMMC plot for. |
Create new vpc theme
Description
Create new vpc theme
Usage
create_vpc_theme(...)
Arguments
... |
pass arguments to 'new_vpc_theme' |
Define data column defaults for various softwares
Description
Define data column defaults for various softwares
Usage
define_data_columns(sim, obs, sim_cols, obs_cols, software_type)
Arguments
sim |
simulated data |
obs |
observed data |
sim_cols |
list for mapping simulation data columns, e.g. 'list(dv = "DV", id = "ID", idv = "TIME", pred="PRED")' |
obs_cols |
list for mapping observation data columns, e.g. 'list(dv = "DV", id = "ID", idv = "TIME", pred="PRED")' |
software_type |
software type, one of 'nonmem', 'phoenix', 'PKPDsim' |
Draw parameters from multivariate distribution
Description
Draw parameters from multivariate distribution
Usage
draw_params_mvr(ids, n_sim, theta, omega_mat, par_names = NULL)
Arguments
ids |
vector of id numbers |
n_sim |
number of simulations |
theta |
theta vector |
omega_mat |
omega matrix |
par_names |
parameter names vector |
Calculate percentiles below / above lloq / uloq
Description
Calculate percentiles below / above lloq / uloq
Usage
loq_perc(x, limit = 1, cens = "left")
Arguments
x |
data |
limit |
censoring limit |
cens |
censoring direction (left/right) |
Create a customized VPC theme
Description
Create a customized VPC theme
Usage
new_vpc_theme(update = NULL)
Arguments
update |
list containing the plot elements to be updated. Run 'new_vpc_theme()' with no arguments to show an overview of available plot elements. |
Details
This function creates a theme that customizes how the VPC looks, i.e. colors, fills, transparencies, linetypes an sizes, etc. The following arguments can be specified in the input list:
obs_color: color for observations points
obs_size: size for observation points
obs_median_color: color for median observation line
obs_median_linetype: linetype for median observation line
obs_median_size: size for median observation line
obs_ci_fill: color for observation CI fill
obs_ci_color: color for observation CI lines
obs_ci_linetype: linetype for observation CI lines
obs_ci_size: size for observations CI lines
sim_pi_fill: fill color for simulated prediction interval areas
sim_pi_alpha: transparency for simulated prediction interval areas
sim_pi_color: color for simulated prediction interval lines
sim_pi_linetype: linetype for simulated prediction interval lines
sim_pi_size: size for simulated prediction interval lines
sim_median_fill: fill color for simulated median area
sim_median_alpha: transparency for simulated median area
sim_median_color: color for simulated median line
sim_median_linetype: linetype for simulated median line
sim_median_size: size for simulated median line
bin_separators_color: color for bin separator lines, NA for don't plot
bin_separators_location: where to plot bin separators ("t" for top, "b" for bottom)
loq_color: color of line showing limit of quantification
Value
A list with vpc theme specifiers
Examples
theme1 <- new_vpc_theme(update = list(
obs_color = "red",
obs_ci_color = "#aa0000",
obs_alpha = .3,
sim_pi_fill = "#cc8833",
sim_pi_size = 2
))
vpc(simple_data$sim, simple_data$obs, vpc_theme = theme1)
Simulate PK data from a 1-compartment iv model
Description
Simulate PK data from a 1-compartment iv model
Usage
pk_iv_1cmt(
t,
t_inf = 1,
tau = 24,
dose = 120,
CL = 0.345,
Vc = 1.75,
ruv = NULL
)
Arguments
t |
Time after dose |
t_inf |
Infusion length |
tau |
Dosing interval |
dose |
Dose |
CL |
Clearance |
Vc |
Volume of distribution |
ruv |
Residual variability |
Value
A vector of predicted values, with or without added residual variability
Examples
dat1 <- vpc:::pk_iv_1cmt(t = c(0:72), tau = 24, dose = 120,
CL = 5, Vc = 50)
dat2 <- vpc:::pk_iv_1cmt(t = c(0:72), tau = 24, dose = 120,
CL = 5, Vc = 50,
ruv = list(proportional = 0.1, additive = 0.1))
Simulate PK data from a 1-compartment oral model
Description
Simulate PK data from a 1-compartment oral model
Usage
pk_oral_1cmt(t, tau = 24, dose = 120, ka = 1, ke = 1, cl = 10, ruv = NULL)
Arguments
t |
Time after dose |
tau |
Dosing interval |
dose |
Dose |
ka |
Absorption rate |
ke |
Elimination rate |
cl |
Clearance |
ruv |
Residual variability |
Value
A vector of predicted values, with or without added residual variability
Examples
dat1 <- vpc:::pk_oral_1cmt(t = c(0:72), tau = 24, dose = 120,
ka = 1, ke = 1, cl = 10)
dat2 <- vpc:::pk_oral_1cmt(t = c(0:72), tau = 24, dose = 120,
ka = 1, ke = 1, cl = 10,
ruv = list(proportional = 0.1, additive = 0.1))
VPC plotting function
Description
This function performs no parsing of data, it just plots the already calculated statistics generated using one of the 'vpc' functions.
Usage
plot_vpc(
db,
show = NULL,
vpc_theme = NULL,
smooth = TRUE,
log_x = FALSE,
log_y = FALSE,
xlab = NULL,
ylab = NULL,
title = NULL,
verbose = FALSE
)
Arguments
db |
object created using the 'vpc' function |
show |
what to show in VPC (obs_dv, obs_ci, pi, pi_as_area, pi_ci, obs_median, sim_median, sim_median_ci) |
vpc_theme |
theme to be used in VPC. Expects list of class vpc_theme created with function vpc_theme() |
smooth |
"smooth" the VPC (connect bin midpoints) or show bins as rectangular boxes. Default is TRUE. |
log_x |
Boolean indicting whether x-axis should be shown as logarithmic. Default is FALSE. |
log_y |
Boolean indicting whether y-axis should be shown as logarithmic. Default is FALSE. |
xlab |
label for x axis |
ylab |
label for y axis |
title |
title |
verbose |
verbosity (T/F) |
See Also
sim_data, vpc_cens, vpc_tte, vpc_cat
Examples
## See vpc.ronkeizer.com for more documentation and examples
library(vpc)
vpc_db <- vpc(sim = simple_data$sim, obs = simple_data$obs, vpcdb = TRUE)
plot_vpc(vpc_db, title = "My new vpc", x = "Custom x label")
Calculate quantiles respecting the censored data
Description
Calculate quantiles respecting the censored data
Usage
quantile_cens(x, p = 0.5, limit = 1, cens = "left")
Arguments
x |
data |
p |
quantile |
limit |
censoring limit |
cens |
censoring direction (left/right) |
NONMEM output table import function
Description
Quickly import NONMEM output tables into R.
Function taken from 'modelviz' package by Benjamin Guiastrennec.
When both skip and header are NULL,
read_nmtab will automatically detect the optimal
settings to import the tables. When more than one files are
provided for a same NONMEM run, they will be combined into
a single data.frame.
Usage
read_table_nm(
file = NULL,
skip = NULL,
header = NULL,
rm_duplicates = FALSE,
nonmem_tab = TRUE
)
Arguments
file |
full file name |
skip |
number of lines to skip before reading data |
header |
logical value indicating whether the file contains the names of the variables as its first line |
rm_duplicates |
logical value indicating whether duplicated columns should be removed |
nonmem_tab |
logical value indicating to the function whether the file is a table or a nonmem additional output file. |
Value
A data.frame
Examples
## Not run:
data <- read_table_nm(file = '../models/pk/sdtab101')
## End(Not run)
Replace list elements by name
Description
Replace list elements by name
Usage
replace_list_elements(list, replacement)
Arguments
list |
original list |
replacement |
replacement list |
Details
Finds and replaces list elements by name and throws an error if an element is not available in the original list. This is a local duplicate of the PKPDmisc copy for the VPC package to reduce dependency on PKPDmisc at this time.
Examples
## Not run:
list <- list(ipred = "ipred", dv = "dv", idv = "idv", "pred" = "pred")
replacement <- list(dv = "conc", idv = "time")
list <- replace_list_elements(list, replacement)
## End(Not run)
Simulated RTTE data (1x)
Description
An example dataset with simulated repeated time-to-event data
Usage
rtte_obs_nm
Format
An object of class data.frame with 573 rows and 6 columns.
Simulated RTTE data (100x)
Description
An example dataset with simulated repeated time-to-event data (100 simulations)
Usage
rtte_sim_nm
Format
An object of class data.frame with 2000000 rows and 7 columns.
Defaults for show argument
Description
Defaults for show argument
Usage
show_default
Format
An object of class list of length 11.
Defaults for show argument for TTE VPC
Description
Defaults for show argument for TTE VPC
Usage
show_default_tte
Format
An object of class list of length 11.
Simulate data based on a model and parameter distributions
Description
Simulate data based on a model and parameter distributions
Usage
sim_data(
design = cbind(id = c(1, 1, 1), idv = c(0, 1, 2)),
model = function(x) { return(x$alpha + x$beta) },
theta,
omega_mat,
par_names,
par_values = NULL,
draw_iiv = "mvrnorm",
error = list(proportional = 0, additive = 0, exponential = 0),
n = 100
)
Arguments
design |
a design dataset. See example |
model |
A function with the first argument the simulation design, i.e. a dataset with the columns ... The second argument to this function is a dataset with parameters for every individual. This can be supplied by the user, or generated by this sim_data if theta and omega_mat are supplied. |
theta |
vector of fixed effect parameters |
omega_mat |
vector of between subject random effects, specified as lower triangle |
par_names |
A character vector linking the parameters in the model to the variables in the dataset. See example. |
par_values |
parameter values |
draw_iiv |
draw between subject random effects? |
error |
see example |
n |
number of simulations to perform |
Details
This function generates the simulated dependent values for use in the VPC plotting function.
Value
a vector of simulated dependent variables (for us in the VPC plotting function)
See Also
A small rich dataset
Description
A small rich dataset
Usage
simple_data
Format
An object of class list of length 2.
Details
a list containing the obs and sim data for an example dataset to run a simple vpc.
Examples
## Not run:
vpc(simple_data$sim, simple_data$obs)
## End(Not run)
Empty ggplot2 theme
Description
Empty ggplot2 theme
Usage
theme_empty()
Examples
vpc(simple_data$sim, simple_data$obs) + theme_empty()
A nicer default theme for ggplot2
Description
A nicer default theme for ggplot2
Usage
theme_plain()
Examples
vpc(simple_data$sim, simple_data$obs) + theme_plain()
Lower to full triangle
Description
Convert the lower triangle of a covariance matrix to a full matrix object
Usage
triangle_to_full(vect)
Arguments
vect |
the lower triangle of a covariance matrix |
VPC function
Description
Creates a VPC plot from observed and simulation data
Usage
vpc(sim, ...)
## Default S3 method:
vpc(sim, ...)
vpc_vpc(
sim = NULL,
obs = NULL,
psn_folder = NULL,
bins = "jenks",
n_bins = "auto",
bin_mid = "mean",
obs_cols = NULL,
sim_cols = NULL,
software = "auto",
show = NULL,
stratify = NULL,
pred_corr = FALSE,
pred_corr_lower_bnd = 0,
pi = c(0.05, 0.95),
ci = c(0.05, 0.95),
uloq = NULL,
lloq = NULL,
log_y = FALSE,
log_y_min = 0.001,
xlab = NULL,
ylab = NULL,
title = NULL,
smooth = TRUE,
vpc_theme = NULL,
facet = "wrap",
scales = "fixed",
labeller = NULL,
vpcdb = FALSE,
verbose = FALSE,
...
)
Arguments
sim |
this is usually a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm. However it can also be an object like a nlmixr or xpose object |
... |
Other arguments sent to other methods (like xpose or nlmixr); Note these arguments are not used in the default vpc and are ignored by the default method. |
obs |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
psn_folder |
instead of specifying "sim" and "obs", specify a PsN-generated VPC-folder |
bins |
either "density", "time", or "data", "none", or one of the approaches available in classInterval() such as "jenks" (default) or "pretty", or a numeric vector specifying the bin separators. |
n_bins |
when using the "auto" binning method, what number of bins to aim for |
bin_mid |
either "mean" for the mean of all timepoints (default) or "middle" to use the average of the bin boundaries. |
obs_cols |
observation dataset column names (list elements: "dv", "idv", "id", "pred") |
sim_cols |
simulation dataset column names (list elements: "dv", "idv", "id", "pred", "sim") |
software |
name of software platform using (e.g. nonmem, phoenix) |
show |
what to show in VPC (obs_dv, obs_ci, pi, pi_as_area, pi_ci, obs_median, sim_median, sim_median_ci) |
stratify |
character vector of stratification variables. Only 1 or 2 stratification variables can be supplied. |
pred_corr |
perform prediction-correction? |
pred_corr_lower_bnd |
lower bound for the prediction-correction |
pi |
simulated prediction interval to plot. Default is c(0.05, 0.95), |
ci |
confidence interval to plot. Default is (0.05, 0.95) |
uloq |
Number or NULL indicating upper limit of quantification. Default is NULL. |
lloq |
Number or NULL indicating lower limit of quantification. Default is NULL. |
log_y |
Boolean indicting whether y-axis should be shown as logarithmic. Default is FALSE. |
log_y_min |
minimal value when using log_y argument. Default is 1e-3. |
xlab |
label for x axis |
ylab |
label for y axis |
title |
title |
smooth |
"smooth" the VPC (connect bin midpoints) or show bins as rectangular boxes. Default is TRUE. |
vpc_theme |
theme to be used in VPC. Expects list of class vpc_theme created with function vpc_theme() |
facet |
either "wrap", "columns", or "rows" |
scales |
either "fixed" (default), "free_y", "free_x" or "free" |
labeller |
ggplot2 labeller function to be passed to underlying ggplot object |
vpcdb |
Boolean whether to return the underlying vpcdb rather than the plot |
verbose |
show debugging information (TRUE or FALSE) |
Value
a list containing calculated VPC information (when vpcdb=TRUE), or a ggplot2 object (default)
See Also
sim_data, vpc_cens, vpc_tte, vpc_cat
Examples
## See vpc.ronkeizer.com for more documentation and examples
library(vpc)
# Basic commands:
vpc(sim = simple_data$sim, obs = simple_data$obs)
vpc(sim = simple_data$sim, obs = simple_data$obs, lloq = 20)
VPC function for categorical
Description
Creates a VPC plot from observed and simulation data for categorical variables.
Usage
vpc_cat(
sim = NULL,
obs = NULL,
psn_folder = NULL,
bins = "jenks",
n_bins = "auto",
bin_mid = "mean",
obs_cols = NULL,
sim_cols = NULL,
software = "auto",
show = NULL,
ci = c(0.05, 0.95),
uloq = NULL,
lloq = NULL,
xlab = NULL,
ylab = NULL,
title = NULL,
smooth = TRUE,
vpc_theme = NULL,
facet = "wrap",
labeller = NULL,
plot = TRUE,
vpcdb = FALSE,
verbose = FALSE
)
Arguments
sim |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
obs |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
psn_folder |
instead of specifying "sim" and "obs", specify a PsN-generated VPC-folder |
bins |
either "density", "time", or "data", "none", or one of the approaches available in classInterval() such as "jenks" (default) or "pretty", or a numeric vector specifying the bin separators. |
n_bins |
when using the "auto" binning method, what number of bins to aim for |
bin_mid |
either "mean" for the mean of all timepoints (default) or "middle" to use the average of the bin boundaries. |
obs_cols |
observation dataset column names (list elements: "dv", "idv", "id", "pred") |
sim_cols |
simulation dataset column names (list elements: "dv", "idv", "id", "pred") |
software |
name of software platform using (e.g. nonmem, phoenix) |
show |
what to show in VPC (obs_ci, pi, pi_as_area, pi_ci, obs_median, sim_median, sim_median_ci) |
ci |
confidence interval to plot. Default is (0.05, 0.95) |
uloq |
Number or NULL indicating upper limit of quantification. Default is NULL. |
lloq |
Number or NULL indicating lower limit of quantification. Default is NULL. |
xlab |
label for x-axis |
ylab |
label for y-axis |
title |
title |
smooth |
"smooth" the VPC (connect bin midpoints) or show bins as rectangular boxes. Default is TRUE. |
vpc_theme |
theme to be used in VPC. Expects list of class vpc_theme created with function vpc_theme() |
facet |
either "wrap", "columns", or "rows" |
labeller |
ggplot2 labeller function to be passed to underlying ggplot object |
plot |
Boolean indicting whether to plot the ggplot2 object after creation. Default is FALSE. |
vpcdb |
boolean whether to return the underlying vpcdb rather than the plot |
verbose |
show debugging information (TRUE or FALSE) |
Value
a list containing calculated VPC information (when vpcdb=TRUE), or a ggplot2 object (default)
See Also
sim_data, vpc, vpc_tte, vpc_cens
Examples
## See vpc.ronkeizer.com for more documentation and examples
library(vpc)
# simple function to simulate categorical data for single individual
sim_id <- function(id = 1) {
n <- 10
logit <- function(x) exp(x) / (1+exp(x))
data.frame(id = id, time = seq(1, n, length.out = n),
dv = round(logit((1:n) - n/2 + rnorm(n, 0, 1.5))) )
}
## simple function to simulate categorical data for a trial
sim_trial <- function(i = 1, n = 20) { # function to simulate categorical data for a trial
data.frame(sim = i, do.call("rbind", lapply(1:n, sim_id)))
}
## simulate single trial for 20 individuals
obs <- sim_trial(n = 20)
## simulate 200 trials of 20 individuals
sim <- do.call("rbind", lapply(1:200, sim_trial, n = 20))
## Plot categorical VPC
vpc_cat(sim = sim, obs = obs)
VPC function for left- or right-censored data (e.g. BLOQ data)
Description
Creates a VPC plot from observed and simulation data for censored data. Function can handle both left- (below lower limit of quantification) and right-censored (above upper limit of quantification) data.
Usage
vpc_cens(
sim = NULL,
obs = NULL,
psn_folder = NULL,
bins = "jenks",
n_bins = 8,
bin_mid = "mean",
obs_cols = NULL,
sim_cols = NULL,
software = "auto",
show = NULL,
stratify = NULL,
stratify_color = NULL,
ci = c(0.05, 0.95),
uloq = NULL,
lloq = NULL,
plot = FALSE,
xlab = "Time",
ylab = "Probability of <LOQ",
title = NULL,
smooth = TRUE,
vpc_theme = NULL,
facet = "wrap",
labeller = NULL,
vpcdb = FALSE,
verbose = FALSE
)
Arguments
sim |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
obs |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
psn_folder |
instead of specifying "sim" and "obs", specify a PsN-generated VPC-folder |
bins |
either "density", "time", or "data", or a numeric vector specifying the bin separators. |
n_bins |
number of bins |
bin_mid |
either "mean" for the mean of all timepoints (default) or "middle" to use the average of the bin boundaries. |
obs_cols |
observation dataset column names (list elements: "dv", "idv", "id", "pred") |
sim_cols |
simulation dataset column names (list elements: "dv", "idv", "id", "pred") |
software |
name of software platform using (e.g. nonmem, phoenix) |
show |
what to show in VPC (obs_ci, pi, pi_as_area, pi_ci, obs_median, sim_median, sim_median_ci) |
stratify |
character vector of stratification variables. Only 1 or 2 stratification variables can be supplied. |
stratify_color |
variable to stratify and color lines for observed data. Only 1 stratification variables can be supplied. |
ci |
confidence interval to plot. Default is (0.05, 0.95) |
uloq |
Number or NULL indicating upper limit of quantification. Default is NULL. |
lloq |
Number or NULL indicating lower limit of quantification. Default is NULL. |
plot |
Boolean indicating whether to plot the ggplot2 object after creation. Default is FALSE. |
xlab |
ylab as numeric vector of size 2 |
ylab |
ylab as numeric vector of size 2 |
title |
title |
smooth |
"smooth" the VPC (connect bin midpoints) or show bins as rectangular boxes. Default is TRUE. |
vpc_theme |
theme to be used in VPC. Expects list of class vpc_theme created with function vpc_theme() |
facet |
either "wrap", "columns", or "rows" |
labeller |
ggplot2 labeller function to be passed to underlying ggplot object |
vpcdb |
boolean whether to return the underlying vpcdb rather than the plot |
verbose |
show debugging information (TRUE or FALSE) |
Value
a list containing calculated VPC information, and a ggplot2 object
See Also
sim_data, vpc, vpc_tte, vpc_cat
Examples
## See vpc.ronkeizer.com for more documentation and examples
library(vpc)
vpc_cens(sim = simple_data$sim, obs = simple_data$obs, lloq = 30)
vpc_cens(sim = simple_data$sim, obs = simple_data$obs, uloq = 120)
VPC function for time-to-event (survival) data
Description
This function can be used for either single time-to-event (TTE) or repeated time-to-event (RTTE) data.
Usage
vpc_tte(
sim = NULL,
obs = NULL,
psn_folder = NULL,
rtte = FALSE,
rtte_calc_diff = TRUE,
rtte_conditional = TRUE,
events = NULL,
bins = FALSE,
n_bins = 10,
software = "auto",
obs_cols = NULL,
sim_cols = NULL,
kmmc = NULL,
reverse_prob = FALSE,
stratify = NULL,
stratify_color = NULL,
ci = c(0.05, 0.95),
plot = FALSE,
xlab = "Time",
ylab = "Survival (%)",
show = NULL,
as_percentage = TRUE,
title = NULL,
smooth = FALSE,
vpc_theme = NULL,
facet = "wrap",
labeller = NULL,
verbose = FALSE,
vpcdb = FALSE
)
Arguments
sim |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
obs |
a data.frame with observed data, containing the independent and dependent variable, a column indicating the individual, and possibly covariates. E.g. load in from NONMEM using read_table_nm |
psn_folder |
instead of specifying "sim" and "obs", specify a PsN-generated VPC-folder |
rtte |
repeated time-to-event data? Default is FALSE (treat as single-event TTE) |
rtte_calc_diff |
recalculate time (T/F)? When simulating in NONMEM, you will probably need to set this to TRUE to recalculate the TIME to relative times between events (unless you output the time difference between events and specify that as independent variable to the vpc_tte() function. |
rtte_conditional |
'TRUE' (default) or 'FALSE'. Compute the probability for each event newly ('TRUE'), or calculate the absolute probability ('FALSE', i.e. the "probability of a 1st, 2nd, 3rd event etc" rather than the "probability of an event happening"). |
events |
numeric vector describing which events to show a VPC for when repeated TTE data, e.g. c(1:4). Default is NULL, which shows all events. |
bins |
either "density", "time", or "data", or a numeric vector specifying the bin separators. |
n_bins |
number of bins |
software |
name of software platform using (e.g. nonmem, phoenix) |
obs_cols |
observation dataset column names (list elements: "dv", "idv", "id", "pred") |
sim_cols |
simulation dataset column names (list elements: "dv", "idv", "id", "pred", "sim") |
kmmc |
either NULL (for regular TTE vpc, default), or a variable name for a KMMC plot (e.g. "WT") |
reverse_prob |
reverse the probability scale (i.e. plot 1-probability) |
stratify |
character vector of stratification variables. Only 1 or 2 stratification variables can be supplied. |
stratify_color |
character vector of stratification variables. Only 1 stratification variable can be supplied, cannot be used in conjunction with 'stratify'. |
ci |
confidence interval to plot. Default is (0.05, 0.95) |
plot |
Boolean indicating whether to plot the ggplot2 object after creation. Default is FALSE. |
xlab |
label for x-axis |
ylab |
label for y-axis |
show |
what to show in VPC (obs_ci, obs_median, sim_median, sim_median_ci) |
as_percentage |
Show y-scale from 0-100 percent? TRUE by default, if FALSE then scale from 0-1. |
title |
title |
smooth |
"smooth" the VPC (connect bin midpoints) or show bins as rectangular boxes. Default is TRUE. |
vpc_theme |
theme to be used in VPC. Expects list of class vpc_theme created with function vpc_theme() |
facet |
either "wrap", "columns", or "rows" |
labeller |
ggplot2 labeller function to be passed to underlying ggplot object |
verbose |
TRUE or FALSE (default) |
vpcdb |
Boolean whether to return the underlying vpcdb rather than the plot |
Details
Creates a VPC plot from observed and simulation survival data
Value
a list containing calculated VPC information, and a ggplot2 object
See Also
sim_data, vpc, vpc_tte, vpc_cens
Examples
## See vpc-docs.ronkeizer.com for more documentation and examples.
## Example for repeated) time-to-event data
## with NONMEM-like data (e.g. simulated using a dense grid)
data(rtte_obs_nm)
data(rtte_sim_nm)
# treat RTTE as TTE, no stratification
vpc_tte(sim = rtte_sim_nm[rtte_sim_nm$sim <= 20,],
obs = rtte_obs_nm,
rtte = FALSE,
sim_cols=list(dv = "dv", idv = "t"), obs_cols=list(idv = "t"))