Type: Package
Title: Clinical Trials with Intercurrent Events Simulator
Version: 0.1.3
Maintainer: Ahmad Hakeem Abdul Wahab <hakeemwahab00@gmail.com>
Description: Simulates clinical trials and summarizes causal effects and treatment policy estimands in the presence of intercurrent events in a transparent and intuitive manner.
URL: https://hakeemwahabapp.shinyapps.io/cities/
License: GPL (≥ 3)
Encoding: UTF-8
RoxygenNote: 7.2.3
VignetteBuilder: knitr
Imports: dplyr, ggplot2, plotly, tidyr, ggthemes
Suggests: testthat (≥ 3.0.0), knitr, rmarkdown
Config/testthat/edition: 3
NeedsCompilation: no
Packaged: 2023-07-27 19:03:41 UTC; aabdulw1
Author: Ahmad Hakeem Abdul Wahab ORCID iD [aut, cre], Yongming Qu [aut], Hege Michiels [aut], Junxiang Luo [aut], Run Zhuang [aut], Dominique McDaniel [aut], Dong Xi [aut], Elena Polverejan [aut], Steven Gilbert [aut], Stephen Ruberg [aut], Arman Sabbagh [aut]
Repository: CRAN
Date/Publication: 2023-08-08 07:50:10 UTC

colSD

Description

Helper function to calculate standard deviation of matrix by columns

Usage

colSD(data_in)

Arguments

data_in

matrix of numeric values

Value

Vector of standard deivations of columns of data_in.

Examples

set.seed(1)
colSD(matrix(rnorm(100), ncol=5))

data_generator

Description

Helper function to simulate single clinical trial

Usage

data_generator(
  n_patient_vector,
  p_loe_max,
  z_l_loe,
  z_u_loe,
  p_ee_max,
  z_l_ee,
  z_u_ee,
  timepoints,
  pacf_list,
  sigma_ar_vec,
  mean_list,
  beta_list,
  p_admin,
  rate_dc_ae,
  prob_ae,
  seed_val,
  reference_id,
  plot_po = FALSE,
  up_good = "Up",
  threshold,
  delta_adjustment_in,
  covariate_df
)

Arguments

n_patient_vector

Vector of number of patients

p_loe_max

The maximum probability of discontinuing due to LoE

z_l_loe

The lower (or left) threshold of the LoE curve

z_u_loe

The upper (or right) threshold of the LoE curve

p_ee_max

The maximum probability of discontinuing due to EE

z_l_ee

The lower (or left) threshold of the EE curve

z_u_ee

The upper (or right) threshold of the EE curve

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

pacf_list

List of pacf vectors

sigma_ar_vec

Vector of variances per arm associated with list of pacf vectors

mean_list

List of vectors of means per arm

beta_list

List of vectors of beta coefficients per arm. All vectors must have the same length and must be the same as the number of columns for the covariate_df.

p_admin

Vector of probabilities of discontinuing due to admin reasons

rate_dc_ae

Vector of probabilities of observing at least one adverse event

prob_ae

Vector of proportions of discontinuing due to adverse event

seed_val

Starting seed value

reference_id

ID for pairwise comparisons, e.g. for three arms, if reference_id=1, then arms 2 and 3 will be compared only to arm 1

plot_po

TRUE, if plotting data only. Otherwise, set to FALSE

up_good

"Up" if higher outcome values indicate better responses

threshold

Value to dichotomize continuous outcomes on

delta_adjustment_in

Vector of delta adjustment values or NA if none. E.g. (2,3,1) when reference_id = 1 means no delta adjustment on arm 1 (even though 2 was supplied, but since arm 1 is the reference arm, this will be defaulted to 0 regardless), 3 on arm 2 and 1 on arm 3.

covariate_df

Matrix or dataframe of covariates. Set NA if using default covariates, which comprises one continuous (standard normal) and binary (bernoulli with prob 0.5) covariates. Rows correspond to the total number of subjects. Order matters. For instance, if you want to simulate a trial with 3 arms, each of size 30,50 and 80, then covariate_df would have 30+50+80 rows such that the first 30 rows are covariates for arm 1, the next 50 rows are covariates for arm 2 and the last 80 rows are covariates for arm 3.

Value

List of dataframes of estimands and simulated data, including delta adjusted ones if requested:

estimand_mean

List of means of the FULL, S_++, S_*+ and PP estimands

estimand_sd

List of standard deviations of the FULL, S_++, S_*+ and PP estimands

dc_mean_list

List of proportions of discontinuations

observed_df

Dataframe of the observed outcomes

po_df

Dataframe of the potential outcomes

ir_df

Dataframe of the outcomes that have been adjusted via immediate reference (IR) or delta adjustment (Delta) for treatment policy estimands. The IR outcomes are labelled as ir_data while the delta adjusted outcomes are labelled as delta_data. The delta adjusted outcomes will only be available if the correct inputs for delta_adjustment_in are provided

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)

n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)

mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)

sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                 c(-0.2, 0.4))

beta_list = list(c(1.25, 1.25),
                 c(1.25, 1.25))
covariate_df = NA

# LoE & EE
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10

# Admin & AE

p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)

prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)

rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)

starting_seed_val = 1
static_output = TRUE

mean_out = plot_means(n_patient_vector = n_patient_vector, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, reference_id = reference_id, seed_val = starting_seed_val,
total_data = total_data, threshold = threshold, covariate_df = covariate_df,
static_output = static_output)

plot_loe_ee (mean_list = mean_list, ref_grp = reference_id,
stdev_vec = sigma_ar_vec, p_loe_max = p_loe_max, z_l_loe = z_l_loe,
z_u_loe = z_u_loe, p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee,
up_good = up_good, greyscale = FALSE, static_output = static_output)

data_out = data_generator(n_patient_vector = n_patient_vector,
p_loe_max = p_loe_max, z_l_loe = z_l_loe, z_u_loe = z_u_loe,
p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, p_admin = p_admin, rate_dc_ae = rate_dc_ae,
prob_ae = prob_ae, seed_val = starting_seed_val, reference_id = reference_id,
plot_po = FALSE, up_good = up_good, threshold = threshold,
delta_adjustment_in = delta_adjustment_in,
covariate_df = covariate_df)

data_generator_loop

Description

Simulate multiple or single clinical trial

Usage

data_generator_loop(
  n_patient_vector,
  p_loe_max,
  z_l_loe,
  z_u_loe,
  p_ee_max,
  z_l_ee,
  z_u_ee,
  timepoints,
  pacf_list,
  sigma_ar_vec,
  mean_list,
  beta_list,
  p_admin,
  rate_dc_ae,
  prob_ae,
  seed_val,
  reference_id,
  plot_po = FALSE,
  up_good,
  threshold,
  total_data,
  delta_adjustment_in,
  covariate_df
)

Arguments

n_patient_vector

Vector of number of patients

p_loe_max

The maximum probability of discontinuing due to LoE

z_l_loe

The lower (or left) threshold of the LoE curve

z_u_loe

The upper (or right) threshold of the LoE curve

p_ee_max

The maximum probability of discontinuing due to EE

z_l_ee

The lower (or left) threshold of the EE curve

z_u_ee

The upper (or right) threshold of the EE curve

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

pacf_list

List of pacf vectors

sigma_ar_vec

Vector of variances per arm associated with list of pacf vectors

mean_list

List of vectors of means per arm

beta_list

List of vectors of beta coefficients per arm. All vectors must have the same length and must be the same as the number of columns for the covariate_df

p_admin

Vector of probabilities of discontinuing due to admin reasons

rate_dc_ae

Vector of probabilities of observing at least one adverse event

prob_ae

Vector of proportions of discontinuing due to adverse event

seed_val

Starting seed value

reference_id

ID for pairwise comparisons, e.g. for three arms, if reference_id=1, then arms 2 and 3 will be compared only to arm 1

plot_po

TRUE, if plotting data only. Otherwise, set to FALSE

up_good

"Up" if higher outcome values indicate better responses

threshold

Value to dichotomize continuous outcomes on

total_data

Total number of clinical trials to simulate

delta_adjustment_in

Vector of delta adjustment values or NA if none. E.g. (2,3,1) when reference_id = 1 means no delta adjustment on arm 1 (even though 2 was supplied, but since arm 1 is the reference arm, this will be defaulted to 0 regardless), 3 on arm 2 and 1 on arm 3.

covariate_df

Matrix or dataframe of covariates. Set NA if using default covariates, which comprises one continuous (standard normal) and binary (bernoulli with prob 0.5) covariates. Rows correspond to the total number of subjects. Order matters. For instance, if you want to simulate a trial with 3 arms, each of size 30,50 and 80, then covariate_df would have 30+50+80 rows such that the first 30 rows are covariates for arm 1, the next 50 rows are covariates for arm 2 and the last 80 rows are covariates for arm 3.

Value

List of dataframes of estimands and simulated data, including delta adjusted ones if requested:

estimand_mean

List of means of the FULL, S_++, S_*+ and PP estimands

estimand_sd

List of standard deviations of the FULL, S_++, S_*+ and PP estimands

dc_mean_list

List of proportions of discontinuations

observed_df

Dataframe of the observed outcomes

po_df

Dataframe of the potential outcomes

ir_df

Dataframe of the outcomes that have been adjusted via immediate reference (IR) or delta adjustment (Delta) for treatment policy estimands. The IR outcomes are labelled as ir_data while the delta adjusted outcomes are labelled as delta_data.The delta adjusted outcomes will only be available if the correct inputs for delta_adjustment_in are provided.

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)

n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)

mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)

sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                 c(-0.2, 0.4))

beta_list = list(c(1.25, 1.25),
                 c(1.25, 1.25))
covariate_df = NA

# LoE & EE
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10

# Admin & AE

p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)

prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)

rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)

starting_seed_val = 1
static_output = TRUE

mean_out = plot_means(n_patient_vector = n_patient_vector, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, reference_id = reference_id, seed_val = starting_seed_val,
total_data = total_data, threshold = threshold, covariate_df = covariate_df,
static_output = static_output)

plot_loe_ee (mean_list = mean_list, ref_grp = reference_id,
stdev_vec = sigma_ar_vec, p_loe_max = p_loe_max, z_l_loe = z_l_loe,
z_u_loe = z_u_loe, p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee,
up_good = up_good, greyscale = FALSE, static_output = static_output)

data_out = data_generator_loop(n_patient_vector = n_patient_vector,
p_loe_max = p_loe_max, z_l_loe = z_l_loe, z_u_loe = z_u_loe,
p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, p_admin = p_admin, rate_dc_ae = rate_dc_ae,
prob_ae = prob_ae, seed_val = starting_seed_val, reference_id = reference_id,
plot_po = FALSE, up_good = up_good, threshold = threshold,
total_data = total_data, delta_adjustment_in = delta_adjustment_in,
covariate_df = covariate_df)

line_parameters

Description

Helper function that returns slope and intercept for line equation using two points in the cartesian plot: (x1, x2) and (y1, y2)

Usage

line_parameters(x1, y1, x2, y2)

Arguments

x1

first value of the point (x1, x2) in the cartesian plot

y1

first value of the point (y1, y2) in the cartesian plot

x2

second value of the point (x1, x2) in the cartesian plot

y2

second value of the point (y1, y2) in the cartesian plot

Value

Vector of slope and intercept for equation of line.

Examples

line_parameters(1,2,4,2)

p_ae_poisson

Description

Helper function that returns probability of discontinuing due to adverse events (AE)

Usage

p_ae_poisson(rate_dc_ae, prob_ae)

Arguments

rate_dc_ae

Probability of observing at least one AE

prob_ae

Proportion of discontinuation due to AE

Value

Probabilities of discontinuing due to AE.

Examples

p_ae_poisson(c(0.9, 0.8), c(0.1, 0.1))

p_loe_ee_function

Description

Helper function that returns probability of discontinuing due to lack of efficacy (LoE) or excess efficacy (EE) via a piecewise linear function

Usage

p_loe_ee_function(z, p_max, z_l, p_min = 0, z_u, up_good = TRUE)

Arguments

z

Vector of numeric values, i.e. change from baseline values

p_max

Maximum probability of discontinuing

z_l

The lower (or left) threshold of the piecewise linear function

p_min

Maximum probability of discontinuing (set to 0)

z_u

The upper (or right) threshold of the piecewise linear function

up_good

TRUE if higher outcome values indicate better responses

Value

Probabilities of discontinuing due to LoE or EE.

Examples

line_parameters(1,2,4,2)

pacf_vec_to_acf

Description

Generate correlation matrix from partial autocorrelations

Usage

pacf_vec_to_acf(pacf_vec, n_repeat)

Arguments

pacf_vec

Vector of partial autocorrelations

n_repeat

number of repeat measures (must be longer than length of pacf_vec)

Value

Correlation matrix from partial autocorrelations.

Examples

pacf_vec_to_acf(c(0.5, -0.1), 5)

plot_dc

Description

Plots the discontinuation rates by timepoints

Usage

plot_dc(
  data_out,
  total_data,
  timepoints,
  normal_output = TRUE,
  static_output = FALSE,
  greyscale = FALSE
)

Arguments

data_out

The output from data_generator_loop()

total_data

Total number of clinical trials to simulate

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

normal_output

TRUE if both plots and numeric values of estimands are requested. FALSE if only plots are requested

static_output

TRUE if static mode requested and FALSE if dynamic plot is requested

greyscale

TRUE if greyscale requested and FALSE for color

Value

Plot and dataframe of proportion of discontinuations.

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)
n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)
mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)
sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                c(-0.2, 0.4))
beta_list = list(c(1.25, 1.25),
                c(1.25, 1.25))
covariate_df = NA
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10
p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)
prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)
rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)
starting_seed_val = 1
static_output = TRUE
data_out = data_generator_loop(n_patient_vector = n_patient_vector,
p_loe_max = p_loe_max, z_l_loe = z_l_loe, z_u_loe = z_u_loe,
p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, p_admin = p_admin, rate_dc_ae = rate_dc_ae,
prob_ae = prob_ae, seed_val = starting_seed_val, reference_id = reference_id,
plot_po = FALSE, up_good = up_good, threshold = threshold,
total_data = total_data, delta_adjustment_in = delta_adjustment_in,
covariate_df = covariate_df)

estimates_out = plot_estimates(data_out = data_out, total_data = total_data,
timepoints = timepoints, reference_id = reference_id, IR_display = IR_display,
normal_output = TRUE, static_output = static_output)

dc_out = plot_dc(data_out = data_out, total_data = total_data,
timepoints = timepoints, static_output = static_output)

plot_estimates

Description

Plots the estimates of the estimands

Usage

plot_estimates(
  data_out,
  total_data,
  timepoints,
  reference_id,
  IR_display = TRUE,
  delta_display = TRUE,
  normal_output = TRUE,
  static_output = FALSE,
  greyscale = FALSE
)

Arguments

data_out

The output from data_generator_loop()

total_data

Total number of clinical trials to simulate

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

reference_id

ID for pairwise comparisons, e.g. for three arms, if reference_id=1, then arms 2 and 3 will be compared only to arm 1

IR_display

TRUE if requested to display Immediate Reference estimand. FALSE otherwise

delta_display

TRUE if requested to display Delta estimand. FALSE otherwise

normal_output

TRUE if both plots and numeric values of estimands are requested. FALSE if only plots are requested

static_output

TRUE if static mode requested and FALSE if dynamic plot is requested

greyscale

TRUE if greyscale requested and FALSE for color

Value

Plot and dataframe of estimands.

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)
delta_display = TRUE
n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)
mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)
sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                c(-0.2, 0.4))
beta_list = list(c(1.25, 1.25),
                c(1.25, 1.25))
covariate_df = NA
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10
p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)
prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)
rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)
starting_seed_val = 1
static_output = TRUE
data_out = data_generator_loop(n_patient_vector = n_patient_vector,
p_loe_max = p_loe_max, z_l_loe = z_l_loe, z_u_loe = z_u_loe,
p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, p_admin = p_admin, rate_dc_ae = rate_dc_ae,
prob_ae = prob_ae, seed_val = starting_seed_val, reference_id = reference_id,
plot_po = FALSE, up_good = up_good, threshold = threshold,
total_data = total_data, delta_adjustment_in = delta_adjustment_in,
covariate_df = covariate_df)

estimates_out = plot_estimates(data_out = data_out, total_data = total_data,
timepoints = timepoints, reference_id = reference_id, IR_display = IR_display,
delta_display = delta_display, normal_output = TRUE, static_output = static_output)

plot_loe_ee

Description

Plots the lack of efficacy (LoE) and excess efficacy (EE) graphs

Usage

plot_loe_ee(
  mean_list,
  ref_grp,
  stdev_vec,
  p_loe_max,
  z_l_loe,
  z_u_loe,
  p_ee_max,
  z_l_ee,
  z_u_ee,
  up_good,
  greyscale,
  static_output = FALSE
)

Arguments

mean_list

List of vectors of means per arm

ref_grp

ID for pairwise comparisons, e.g. for three arms, if reference_id=1, then arms 2 and 3 will be compared only to arm 1

stdev_vec

Vector of standard deviations per arm. This is used to adjust the x-axis for display

p_loe_max

The maximum probability of discontinuing due to LoE

z_l_loe

The lower (or left) threshold of the LoE curve

z_u_loe

The upper (or right) threshold of the LoE curve

p_ee_max

The maximum probability of discontinuing due to EE

z_l_ee

The lower (or left) threshold of the EE curve

z_u_ee

The upper (or right) threshold of the EE curve

up_good

"Up" if higher outcome values indicate better responses and "Down" otherwise

greyscale

TRUE for greyscale setting and FALSE for color setting

static_output

TRUE, if static and FALSE if dynamic plot is requested

Value

The plot for LoE and EE.

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)

n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)

mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)

sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                 c(-0.2, 0.4))

beta_list = list(c(1.25, 1.25),
                 c(1.25, 1.25))
covariate_df = NA

# LoE & EE
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10

# Admin & AE

p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)

prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)

rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)

starting_seed_val = 1
static_output = TRUE

mean_out = plot_means(n_patient_vector = n_patient_vector, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, reference_id = reference_id, seed_val = starting_seed_val,
total_data = total_data, threshold = threshold, covariate_df = covariate_df,
static_output = static_output)

plot_loe_ee (mean_list = mean_list, ref_grp = reference_id,
stdev_vec = sigma_ar_vec, p_loe_max = p_loe_max, z_l_loe = z_l_loe,
z_u_loe = z_u_loe, p_ee_max = p_ee_max, z_l_ee = z_l_ee, z_u_ee = z_u_ee,
up_good = up_good, greyscale = FALSE, static_output = static_output)

plot_means

Description

Plots the means of simulation parameters.

Usage

plot_means(
  n_patient_vector,
  timepoints,
  pacf_list,
  sigma_ar_vec,
  mean_list,
  beta_list,
  reference_id,
  seed_val,
  threshold,
  total_data,
  covariate_df,
  static_output = FALSE
)

Arguments

n_patient_vector

Vector of number of patients

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

pacf_list

List of pacf vectors

sigma_ar_vec

Vector of variances per arm associated with list of pacf vectors

mean_list

List of vectors of means per arm

beta_list

List of vectors of beta coefficients per arm. All vectors must have the same length and must be the same as the number of columns for the covariate_df.

reference_id

ID for pairwise comparisons, e.g. for three arms, if reference_id=1, then arms 2 and 3 will be compared only to arm 1

seed_val

Starting seed value

threshold

Value to dichotomize continuous outcomes on

total_data

Total number of clinical trials to simulate

covariate_df

Matrix or dataframe of covariates. Rows correspond to the total number of subjects. Order matters, For instance, if you want to simulate a trial with 3 arms, each of size 30,50 and 80, then covariate_df would have 30+50+80 rows such that the first 30 rows are covariates for arm 1, the next 50 rows are covariates for arm 2 and the last 80 rows are covariates for arm 3.

static_output

TRUE, if static and FALSE if dynamic plot is requested

Value

The plot of raw means.

Examples

total_data = 3
reference_id = 1
threshold = NA
timepoints = c(0,24,48,72,96,120,144)
IR_display = TRUE
delta_adjustment_in = c(0,1)

n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_total = sum(n_patient_vector)

mean_control = c(0,0,0,0,0,0,0)
mean_treatment = c(0,0.1,0.2,0.4,0.6,0.8,1)
mean_list = list(mean_control, mean_treatment)

sigma_ar_vec = c(1, 1)
pacf_list = list(c(-0.2, 0.4),
                 c(-0.2, 0.4))

beta_list = list(c(1.25, 1.25),
                 c(1.25, 1.25))
covariate_df = NA

# LoE & EE
up_good = "Up"
p_loe_max = 0.75
z_l_loe = -7
z_u_loe = -1
p_ee_max = 0.1
z_l_ee = 4
z_u_ee = 10

# Admin & AE

p_admin_ctrl = 0.02
p_admin_expt = 0.02
p_admin = c(p_admin_ctrl, p_admin_expt)

prob_ae_ctrl = 0.7
prob_ae_expt = 0.9
prob_ae = c(prob_ae_ctrl, prob_ae_expt)

rate_dc_ae_ctrl = 0.1
rate_dc_ae_expt = 0.1
rate_dc_ae = c(rate_dc_ae_ctrl, rate_dc_ae_expt)

starting_seed_val = 1
static_output = TRUE

mean_out = plot_means(n_patient_vector = n_patient_vector, timepoints = timepoints,
pacf_list = pacf_list, sigma_ar_vec = sigma_ar_vec, mean_list = mean_list,
beta_list = beta_list, reference_id = reference_id, seed_val = starting_seed_val,
total_data = total_data, threshold = threshold, covariate_df = covariate_df,
static_output = static_output)

rep_rcol

Description

Helper function to repeat a matrix by column

Usage

rep_col(x, n)

Arguments

x

vector to repeat

n

number of repetions

Value

matrix with vector x repeated n-times by columns.

Examples

set.seed(1)
rep_col(rnorm(5), 5)

rep_row

Description

Helper function to repeat a matrix by row

Usage

rep_row(x, n)

Arguments

x

vector to repeat

n

number of repetions

Value

Matrix with vector x repeated n-times by rows.

Examples

set.seed(1)
rep_row(rnorm(5), 5)

simulated_data_output

Description

Helper function to combine simulated data

Usage

simulated_data_output(
  n_patient_cumsum,
  i,
  first_patient,
  data_in,
  covariate_df,
  timepoints,
  beta_list,
  seed_val,
  potential_outcomes = FALSE,
  observed_indicator = NA
)

Arguments

n_patient_cumsum

Vector of number of patients

i

Index for arm

first_patient

Index for first patient of arm

data_in

Simulated data from data_generator()

covariate_df

Matrix or dataframe of covariates. Rows correspond to the total number of subjects. Order matters, For instance, if you want to simulate a trial with 3 arms, each of size 30,50 and 80, then covariate_df would have 30+50+80 rows such that the first 30 rows are covariates for arm 1, the next 50 rows are covariates for arm 2 and the last 80 rows are covariates for arm 3.

timepoints

Vector of timepoints (e.g. weeks, days, time indices)

beta_list

List of vectors of beta coefficients per arm. All vectors must have the same length and must be the same as the number of columns for the covariate_df.

seed_val

Current seed value

potential_outcomes

TRUE if data to be combined is for potential outcomes, and FALSE otherwise

observed_indicator

Dataframe containing which subjects/arms/timepoints were observed (necessary for potential outcomes), else default to NA

Value

Dataframe of for either potential outcomes, observed outcomes, outcomes with immediate reference assumption or delta adjustment assumption

Examples

n_patient_ctrl = 120
n_patient_expt = 150
n_patient_vector = c(n_patient_ctrl, n_patient_expt)
n_patient_cumsum = cumsum(n_patient_vector)
total_patients = sum(n_patient_vector)
timepoints = c(0,24,48,72,96,120,144)
data_in = matrix(rnorm(length(timepoints)*n_patient_ctrl), ncol = length(timepoints))
i = 1
first_patient = 1
covariate_df = data.frame(continuous  = rnorm(n = total_patients, mean = 0, sd = 1),
binary = rbinom(n = total_patients, size = 1, prob = 0.5))
beta_list = NA
seed_val = 1
potential_outcomes = FALSE
observed_indicator = NA
simulated_data_output(n_patient_cumsum = n_patient_cumsum, i = i,
first_patient = first_patient, data_in = data_in, covariate_df = covariate_df,
timepoints = timepoints, beta_list = beta_list, seed_val = seed_val,
potential_outcomes = FALSE, observed_indicator = NA)