simtrial

https://merck.github.io/simtrial/
Admin
simtrial is a clinical trial simulation framework for time-to-event endpoints, supporting fixed,...
Tokens:100,880
Snippets:551
Trust Score:9.7
Update:1 month ago
Show doc for...
Tokens:
Raw
### Example: load dplyr and future packages

Source: https://merck.github.io/simtrial/reference/sim_fixed_n.html

The example section begins by loading the dplyr and future packages, which are used in subsequent simulation examples.

```r
library(dplyr)
library(future)
```

--------------------------------

### Example: Setting up trial parameters for get_analysis_date

Source: https://merck.github.io/simtrial/reference/get_analysis_date.html

Initializes parameters such as alpha, ratio, sample size, and study duration for use with the analysis date function. This example is from the documentation's Examples section.

```r
library(gsDesign2)

alpha <- 0.025
ratio <- 3
n <- 500
info_frac <- c(0.7, 1)
prevalence_ratio <- c(0.4, 0.6)
study_duration <- 48
```

--------------------------------

### Example: Simulate data and get cut date for 50 events

Source: https://merck.github.io/simtrial/reference/get_cut_date_by_event.html

Complete example that simulates a time-to-event dataset with `sim_pw_surv()`, filters to the 'Positive' stratum, gets the cut date for 50 events, cuts the data, and displays the event table. The output shows the resulting counts.

```r
library(dplyr)
#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union

# Use default enrollment and calendar cut date
# for 50 events in the "Positive" stratum
x <- sim_pw_surv(
  n = 200,
  stratum = data.frame(
    stratum = c("Positive", "Negative"),
    p = c(.5, .5)
  ),
  fail_rate = data.frame(
    stratum = rep(c("Positive", "Negative"), 2),
    period = rep(1, 4),
    treatment = c(rep("control", 2), rep("experimental", 2)),
    duration = rep(1, 4),
    rate = log(2) / c(6, 9, 9, 12)
  ),
  dropout_rate = data.frame(
    stratum = rep(c("Positive", "Negative"), 2),
    period = rep(1, 4),
    treatment = c(rep("control", 2), rep("experimental", 2)),
    duration = rep(1, 4),
    rate = rep(.001, 4)
  )
)

d <- get_cut_date_by_event(x |> filter(stratum == "Positive"), event = 50)

y <- cut_data_by_date(x, cut_date = d)
table(y$stratum, y$event)
#>           
#>             0  1
#>   Negative 49 45
#>   Positive 37 50
```

--------------------------------

### Install simtrial from CRAN

Source: https://merck.github.io/simtrial/index.html

Install the stable version of simtrial from CRAN.

```r
install.packages("simtrial")
```

--------------------------------

### Install simtrial development version from GitHub

Source: https://merck.github.io/simtrial/index.html

Install the development version of simtrial from GitHub to access new features or bug fixes. Requires the remotes package.

```r
# install.packages("remotes")
remotes::install_github("Merck/simtrial")
```

--------------------------------

### Example: Summarize group sequential simulations with design and bound

Source: https://merck.github.io/simtrial/reference/summary.html

Complete example showing how to set up enrollment and failure rates, build a one-sided group sequential design with gs_design_ahr(), run simulations with sim_gs_n(), and summarize results with and without the design. The output shows the summary data frames.

```r
library(gsDesign2)

# Parameters for enrollment
enroll_rampup_duration <- 4 # Duration for enrollment ramp up
enroll_duration <- 16 # Total enrollment duration
enroll_rate <- define_enroll_rate(
  duration = c(
    enroll_rampup_duration, enroll_duration - enroll_rampup_duration),
 rate = c(10, 30))

# Parameters for treatment effect
delay_effect_duration <- 3 # Delay treatment effect in months
median_ctrl <- 9 # Survival median of the control arm
median_exp <- c(9, 14) # Survival median of the experimental arm
dropout_rate <- 0.001
fail_rate <- define_fail_rate(
  duration = c(delay_effect_duration, 100),
  fail_rate = log(2) / median_ctrl,
  hr = median_ctrl / median_exp,
  dropout_rate = dropout_rate)

# Other related parameters
alpha <- 0.025 # Type I error
beta <- 0.1 # Type II error
ratio <- 1 # Randomization ratio (experimental:control)

# Build a one-sided group sequential design
design <- gs_design_ahr(
  enroll_rate = enroll_rate, fail_rate = fail_rate,
  ratio = ratio, alpha = alpha, beta = beta,
  analysis_time = c(12, 24, 36),
  upper = gs_spending_bound,
  upar = list(sf = gsDesign::sfLDOF, total_spend = alpha),
  lower = gs_b,
  lpar = rep(-Inf, 3))

# Define cuttings of 2 IAs and 1 FA
ia1_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[1]))
ia2_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[2]))
fa_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[3]))

# Run simulations
simulation <- sim_gs_n(
  n_sim = 3,
  sample_size = ceiling(design$analysis$n[3]),
  enroll_rate = design$enroll_rate,
  fail_rate = design$fail_rate,
  test = wlr,
  cut = list(ia1 = ia1_cut, ia2 = ia2_cut, fa = fa_cut),
  weight = fh(rho = 0, gamma = 0.5))
#> Backend uses sequential processing.

# Summarize simulations
bound <- gsDesign::gsDesign(k = 3, test.type = 1, sfu = gsDesign::sfLDOF)$upper$bound
simulation |> summary(bound = bound)
#>   analysis    sim_n sim_event sim_time sim_upper_prob
#> 1        1 353.6667        97 12.29699             NA
#> 2        2 505.0000       305 23.90850      0.6666667
#> 3        3 505.0000       405 35.79231      1.0000000

# Summarize simulation and compare with the planned design
simulation |> summary(design = design)
#>   analysis asy_upper_prob sim_upper_prob sim_event    sim_n sim_time asy_time
#> 1        1   0.0001486592             NA        97 353.6667 12.29699       12
#> 2        2   0.5723210881      0.6666667       305 505.0000 23.90850       24
#> 3        3   0.8999997572      1.0000000       405 505.0000 35.79231       36
#>      asy_n asy_event
#> 1 353.0464  96.77449
#> 2 504.3520 304.00970
#> 3 504.3520 404.14162
```

--------------------------------

### Load required R packages for simtrial discrepancy example

Source: https://merck.github.io/simtrial/articles/discrepancy-between-simtrial-and-survival.html

Loads the packages used in the article: gsDesign, gsDesign2, dplyr, gt, simtrial, tidyr, and survival. Required before running the simulation and comparison examples.

```r
library(gsDesign)
library(gsDesign2)
library(dplyr)
library(gt)
library(simtrial)
library(tidyr)
library(survival)
```

--------------------------------

### Install bshazard package from GitHub

Source: https://merck.github.io/simtrial/articles/arbitrary-hazard.html

Installs the bshazard package from GitHub if it is not available on CRAN. Run this before using bshazard::bshazard() later in the vignette.

```r
remotes::install_github("cran/bshazard")
```

--------------------------------

### Example 1: counting_process on a small data frame

Source: https://merck.github.io/simtrial/reference/counting_process.html

Example 1 creates a small data frame with two strata and calls counting_process with arm = 1. The output shows the counting process dataset with columns including stratum, event_total, event_trt, tte, n_risk_total, n_risk_trt, s, o_minus_e, and var_o_minus_e.

```r
# Example 1
x <- data.frame(
  stratum = c(rep(1, 10), rep(2, 6)),
  treatment = rep(c(1, 1, 0, 0), 4),
  tte = 1:16,
  event = rep(c(0, 1), 8)
)
counting_process(x, arm = 1)
#>   stratum event_total event_trt tte n_risk_total n_risk_trt         s
#> 1       1           1         1   2            9          5 1.0000000
#> 2       1           1         0   4            7          4 0.8888889
#> 3       1           1         1   6            5          3 0.7619048
#> 4       1           1         0   8            3          2 0.6095238
#> 5       2           1         0  12            5          2 1.0000000
#> 6       2           1         1  14            3          1 0.8000000
#>    o_minus_e var_o_minus_e
#> 1  0.4444444     0.2469136
#> 2 -0.5714286     0.2448980
#> 3  0.4000000     0.2400000
#> 4 -0.6666667     0.2222222
#> 5 -0.4000000     0.2400000
#> 6  0.6666667     0.2222222
```

--------------------------------

### Exponential and piecewise exponential failure time simulation with rpwexp()

Source: https://merck.github.io/simtrial/reference/rpwexp.html

Two example workflows cover a constant failure rate (Example 1) and a three-interval piecewise rate (Example 2), generating 10000 failure times and plotting the empirical survival curve on a log scale.

```R
# Example 1
# Exponential failure times
x <- rpwexp(
  n = 10000,
  fail_rate = data.frame(rate = 5, duration = 1)
)
plot(sort(x), (10000:1) / 10001,
  log = "y", main = "Exponential simulated survival curve",
  xlab = "Time", ylab = "P{Survival}"
)

```

```R

# Example 2

# Get 10k piecewise exponential failure times.
# Failure rates are 1 for time 0 to 0.5, 3 for time 0.5 to 1, and 10 for > 1.
# Intervals specifies duration of each failure rate interval
# with the final interval running to infinity.
x <- rpwexp(
  n = 1e4,
  fail_rate = data.frame(rate = c(1, 3, 10), duration = c(.5, .5, 1))
)
plot(sort(x), (1e4:1) / 10001,
  log = "y", main = "PW Exponential simulated survival curve",
  xlab = "Time", ylab = "P{Survival}"
)

```

--------------------------------

### Switch to sequential backend with plan(sequential)

Source: https://merck.github.io/simtrial/articles/parallel.html

Switches the future backend back to sequential execution. Use this to revert from the multisession backend used in the parallel example.

```r
plan(sequential)
```

--------------------------------

### Example 2: counting_process with simulated data and weighted logrank test

Source: https://merck.github.io/simtrial/reference/counting_process.html

Example 2 simulates a survival dataset with sim_pw_surv, cuts it at 150 events, and applies counting_process with arm = "experimental". It then computes the weighted logrank test Z-value and one-sided p-value from the o_minus_e and var_o_minus_e columns.

```r
# Example 2
x <- sim_pw_surv(n = 400)
y <- cut_data_by_event(x, 150) |> counting_process(arm = "experimental")
# Weighted logrank test (Z-value and 1-sided p-value)
z <- sum(y$o_minus_e) / sqrt(sum(y$var_o_minus_e))
c(z, pnorm(z))
#> [1] -3.808578e+00  6.988404e-05
```

--------------------------------

### milestone() example with naive test

Source: https://merck.github.io/simtrial/reference/milestone.html

Example workflow that simulates survival data, cuts it at 150 events, and applies the milestone() function with test_type = "naive". The output shows the method, parameter, estimate, standard error, and z statistic.

```r
cut_data <- sim_pw_surv(n = 200) |>
  cut_data_by_event(150)

cut_data |>
  milestone(10, test_type = "naive")
#> $method
#> [1] "milestone"
#> 
#> $parameter
#> [1] 10
#> 
#> $estimate
#> [1] 0.1760742
#> 
#> $se
#> [1] 0.06979553
#> 
#> $z
#> [1] 2.522714
#> 
```

--------------------------------

### Load required packages

Source: https://merck.github.io/simtrial/articles/modest-wlrt.html

Loads the simtrial, dplyr, and survival packages required for the examples on this page.

```r
library(simtrial)
library(dplyr)
library(survival)
```

--------------------------------

### rmst() with formula interface

Source: https://merck.github.io/simtrial/reference/rmst.html

Example using the formula interface with Surv(month, evntd) ~ trt. The output is identical to the string-column example, confirming the formula is equivalent.

```r
# Formula interface
rmst(
  data = ex1_delayed_effect,
  formula = Surv(month, evntd) ~ trt,
  tau = 10,
  reference = "0"
)
#> $method
#> [1] "RMST"
#> 
#> $parameter
#> [1] 10
#> 
#> $estimate
#> [1] 0.8650493
#> 
#> $se
#> [1] 0.3900344
#> 
#> $z
#> [1] 2.21788
#> 
```

--------------------------------

### milestone() example with log-log test

Source: https://merck.github.io/simtrial/reference/milestone.html

Example workflow that simulates survival data, cuts it at 150 events, and applies the milestone() function with test_type = "log-log". The output shows the method, parameter, estimate, standard error, and z statistic.

```r
cut_data <- sim_pw_surv(n = 200) |>
  cut_data_by_event(150)

cut_data |>
  milestone(10, test_type = "log-log")
#> $method
#> [1] "milestone"
#> 
#> $parameter
#> [1] 10
#> 
#> $estimate
#> [1] 0.5293995
#> 
#> $se
#> [1] 0.2147341
#> 
#> $z
#> [1] 2.465372
#> 
```

--------------------------------

### fit_pwexp() examples with ex1_delayed_effect and treatment comparison

Source: https://merck.github.io/simtrial/reference/fit_pwexp.html

Runs fit_pwexp() on the ex1_delayed_effect dataset, first with default arguments and then separately by treatment group. The output shows the estimated rates and -2 log-likelihood per interval, and the code computes a chi-square test for treatment effect and compares with a logrank test via survdiff(). Note that the example uses rate0 and rate1 objects that are not defined in the visible code, so the final pchisq() call may fail if run as-is.

```r
# Use default arguments for delayed effect example dataset (ex1_delayed_effect)
library(survival)

# Example 1
rateall <- fit_pwexp()
rateall
#>   intervals     ttot event       rate     m2ll
#> 1         3 937.1785    97 0.10350216 634.0236
#> 2         3 605.3572    71 0.11728612 446.3257
#> 3         3 346.8482    30 0.08649317 206.8614
#> 4       Inf 254.1148    20 0.07870458 141.6822

# Example 2
# Estimate by treatment effect
rate1 <- with(subset(ex1_delayed_effect, trt == 1), fit_pwexp(Surv(month, evntd)))
rate0 <- with(subset(ex1_delayed_effect, trt == 0), fit_pwexp(Surv(month, evntd)))

rate1
#>   intervals     ttot event       rate      m2ll
#> 1         3 620.4375    64 0.10315302 418.75734
#> 2         3 415.8482    36 0.08657005 248.16970
#> 3         3 256.2053    19 0.07415927 136.85853
#> 4       Inf 205.4186    13 0.06328542  97.76261
rate0
#>   intervals      ttot event      rate      m2ll
#> 1         3 316.74106    33 0.1041861 215.26408
#> 2         3 189.50899    35 0.1846878 188.23619
#> 3         3  90.64288    11 0.1213554  68.39871
#> 4       Inf  48.69624     7 0.1437483  41.15568
rate1$rate / rate0$rate
#> [1] 0.9900847 0.4687372 0.6110917 0.4402517

# Chi-square test for (any) treatment effect (8 - 4 parameters = 4 df)
pchisq(sum(rateall$m2ll) - sum(rate1$m2ll + rate0$m2ll),
  df = 4,
  lower.tail = FALSE
)
#> [1] 0.006424744

# Compare with logrank
survdiff(formula = Surv(month, evntd) ~ trt, data = ex1_delayed_effect)
#> Call:
#> survdiff(formula = Surv(month, evntd) ~ trt, data = ex1_delayed_effect)
#> 
#>         N Observed Expected (O-E)^2/E (O-E)^2/V
#> trt=0 121       86     67.7      4.97      7.35
#> trt=1 240      132    150.3      2.24      7.35
#> 
#>  Chisq= 7.3  on 1 degrees of freedom, p= 0.007 

# Example 3
# Simple model with 3 rates same for each for 3 months,
# different for each treatment after months
rate1a <- with(subset(ex1_delayed_effect, trt == 1), fit_pwexp(Surv(month, evntd), 3))
rate0a <- with(subset(ex1_delayed_effect, trt == 0), fit_pwexp(Surv(month, evntd), 3))
rate1a$rate / rate0a$rate
#> [1] 0.9900847 0.4808339

m2ll0 <- rateall$m2ll[1] + rate1a$m2ll[2] + rate0a$m2ll[2]
m2ll1 <- sum(rate0$m2ll) + sum(rate1$m2ll)

# As a measure of strength, chi-square examines improvement in likelihood
pchisq(m2ll0 - m2ll1, df = 5, lower.tail = FALSE)
#> [1] 0.741822
```

--------------------------------

### Load required libraries for simtrial MaxCombo vignette

Source: https://merck.github.io/simtrial/articles/maxcombo.html

Loads the simtrial, knitr, dplyr, and gt packages needed for the examples in this vignette.

```r
library(simtrial)
library(knitr)
library(dplyr)
library(gt)
```

--------------------------------

### maxcombo() example with return_corr = TRUE

Source: https://merck.github.io/simtrial/reference/maxcombo.html

Example pipeline that simulates a survival dataset, cuts it at 150 events, and applies `maxcombo()` with `rho = c(0, 0)` and `gamma = c(0, 1)`. The `return_corr = TRUE` flag adds the correlation matrix to the output. The displayed output shows the method, parameter, z-scores, correlation matrix, and p-value.

```r
sim_pw_surv(n = 200) |>
  cut_data_by_event(150) |>
  maxcombo(rho = c(0, 0), gamma = c(0, 1), return_corr = TRUE)
#> $method
#> [1] "MaxCombo"
#> 
#> $parameter
#> [1] "FH(0, 0) + FH(0, 1)"
#> 
#> $z
#> [1] -1.464343 -1.819921
#> 
#> $corr
#>          v1        v2
#> 1 1.0000000 0.8574337
#> 2 0.8574337 1.0000000
#> 
#> $p_value
#> [1] 0.05015051
#> 
```

--------------------------------

### Create and apply a cutting function with create_cut()

Source: https://merck.github.io/simtrial/reference/create_cut.html

Example from the documentation showing how to simulate trial data with sim_pw_surv(), create a cutting function with create_cut() using planned_calendar_time = 45 and target_event_overall = 350, then apply it to the trial data. The output is a cut date of 79.83332.

```r
# Simulate trial data
trial_data <- sim_pw_surv()

# Create a cutting function that applies the following 2 conditions:
# - At least 45 months have passed since the start of the study
# - At least 300 events have occurred
cutting <- create_cut(
  planned_calendar_time = 45,
  target_event_overall = 350
)

# Cut the trial data
cutting(trial_data)
#> [1] 79.83332
```

--------------------------------

### Run multitest() with WLR, RMST, and MaxCombo tests

Source: https://merck.github.io/simtrial/reference/multitest.html

Example from the multitest() documentation. It simulates trial data, cuts it by event count, creates three test functions with create_test(), and runs multitest() with named arguments. The displayed output shows the returned list structure with method, parameter, estimate, se, z, info, info0, and p_value fields.

```r
trial_data <- sim_pw_surv(n = 200)
trial_data_cut <- cut_data_by_event(trial_data, 150)

# create cutting test functions
wlr_partial <- create_test(wlr, weight = fh(rho = 0, gamma = 0))
rmst_partial <- create_test(rmst, tau = 20)
maxcombo_partial <- create_test(maxcombo, rho = c(0, 0), gamma = c(0, 0.5))

multitest(
  data = trial_data_cut,
  wlr = wlr_partial,
  rmst = rmst_partial,
  maxcombo = maxcombo_partial
)
#> $wlr
#> $wlr$method
#> [1] "WLR"
#> 
#> $wlr$parameter
#> [1] "FH(rho=0, gamma=0)"
#> 
#> $wlr$estimate
#> [1] -18.81751
#> 
#> $wlr$se
#> [1] 5.97492
#> 
#> $wlr$z
#> [1] 3.149416
#> 
#> $wlr$info
#> [1] 37.17333
#> 
#> $wlr$info0
#> [1] 37.5
#> 
#> 
#> $rmst
#> $rmst$method
#> [1] "RMST"
#> 
#> $rmst$parameter
#> [1] 20
#> 
#> $rmst$estimate
#> [1] 2.948586
#> 
#> $rmst$se
#> [1] 1.063905
#> 
#> $rmst$z
#> [1] 2.771474
#> 
#> 
#> $maxcombo
#> $maxcombo$method
#> [1] "MaxCombo"
#> 
#> $maxcombo$parameter
#> [1] "FH(0, 0) + FH(0, 0.5)"
#> 
#> $maxcombo$z
#> [1] -3.149416 -3.268623
#> 
#> $maxcombo$p_value
#> [1] 0.0007944802
#> 
#> 
```

--------------------------------

### Study parameters for Freidlin-Korn example

Source: https://merck.github.io/simtrial/articles/modest-wlrt.html

Specifies study parameters for the Freidlin and Korn example: 5-year study duration, 2000 patients, very short enrollment period, and piecewise failure rates with hazard ratios 16 for 0.1 years and 0.76 thereafter.

```r
study_duration <- 5
sample_size <- 2000
enroll_duration <- .0001
enroll_rate <- data.frame(
  duration = enroll_duration,
  rate = sample_size / enroll_duration
)
fail_rate <- data.frame(
  stratum = "All",
  fail_rate = 0.25,
  dropout_rate = 0,
  hr = c(4 / .25, .19 / .25),
  duration = c(.1, 4.9)
)
```

--------------------------------

### Run simulation with nested parallel backend using foreach and %dofuture%

Source: https://merck.github.io/simtrial/articles/parallel.html

Runs the simulation using the nested parallel backend. Sets a seed, defines enrollment rates, and uses foreach with %dofuture% to execute sim_fixed_n in parallel. The .options.future argument with seed = TRUE ensures reproducibility.

```R
set.seed(1)

enroll_rates <- list(enroll_rate1, enroll_rate2)

nested_result <- foreach::foreach(
  i = 1:2,
  .combine = "list",
  .options.future = list(seed = TRUE)
) %dofuture% {
  sim_fixed_n(
    n_sim = n_sim,
    sample_size = 3000,
    target_event = 700,
    enroll_rate = enroll_rates[[i]],
    timing_type = 2 # Time until targeted event count achieved
  )
}
```

--------------------------------

### Summarize simulations and convert to gt table

Source: https://merck.github.io/simtrial/reference/as_gt.html

Complete example that sets up enrollment and failure rates, builds a group sequential design, runs simulations, summarizes results with a bound, and converts to a gt table. The output shows the resulting table with columns for analysis, time, N, event, and crossing probability.

```r
# Parameters for enrollment
enroll_rampup_duration <- 4 # Duration for enrollment ramp up
enroll_duration <- 16 # Total enrollment duration
enroll_rate <- gsDesign2::define_enroll_rate(
  duration = c(
    enroll_rampup_duration, enroll_duration - enroll_rampup_duration),
 rate = c(10, 30))

# Parameters for treatment effect
delay_effect_duration <- 3 # Delay treatment effect in months
median_ctrl <- 9 # Survival median of the control arm
median_exp <- c(9, 14) # Survival median of the experimental arm
dropout_rate <- 0.001
fail_rate <- gsDesign2::define_fail_rate(
  duration = c(delay_effect_duration, 100),
  fail_rate = log(2) / median_ctrl,
  hr = median_ctrl / median_exp,
  dropout_rate = dropout_rate)

# Other related parameters
alpha <- 0.025 # Type I error
beta <- 0.1 # Type II error
ratio <- 1 # Randomization ratio (experimental:control)

# Build a one-sided group sequential design
design <- gsDesign2::gs_design_ahr(
  enroll_rate = enroll_rate, fail_rate = fail_rate,
  ratio = ratio, alpha = alpha, beta = beta,
  analysis_time = c(12, 24, 36),
  upper = gsDesign2::gs_spending_bound,
  upar = list(sf = gsDesign::sfLDOF, total_spend = alpha),
  lower = gsDesign2::gs_b,
  lpar = rep(-Inf, 3))

# Define cuttings of 2 IAs and 1 FA
ia1_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[1]))
ia2_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[2]))
fa_cut <- create_cut(target_event_overall = ceiling(design$analysis$event[3]))

# Run simulations
simulation <- sim_gs_n(
  n_sim = 3,
  sample_size = ceiling(design$analysis$n[3]),
  enroll_rate = design$enroll_rate,
  fail_rate = design$fail_rate,
  test = wlr,
  cut = list(ia1 = ia1_cut, ia2 = ia2_cut, fa = fa_cut),
  weight = fh(rho = 0, gamma = 0.5))
#> Backend uses sequential processing.

# Summarize simulations
simulation |>
 summary(bound = gsDesign::gsDesign(k = 3, test.type = 1, sfu = gsDesign::sfLDOF)$upper$bound) |>
 simtrial::as_gt()
```

--------------------------------

### rmst_single_arm example with ex1_delayed_effect

Source: https://merck.github.io/simtrial/reference/rmst_single_arm.html

Example usage of rmst_single_arm with the ex1_delayed_effect dataset, filtering to treatment arm 1. The call computes RMST at tau=10 for the treatment group, with alpha=0.05. The output shows the resulting data frame with cutoff_time=10, group="Treatment 1", rmst=6.495175, variance=0.05711322, std=0.2389837, lcl=6.026776, ucl=6.963575, and event=127.

```r
data(ex1_delayed_effect)
data_single_arm <- ex1_delayed_effect[ex1_delayed_effect$trt == 1, ]
simtrial:::rmst_single_arm(
  time_var = data_single_arm$month,
  event_var = data_single_arm$evntd,
  tau = 10,
  group_label = "Treatment 1",
  alpha = 0.05
)
#>   cutoff_time       group     rmst   variance       std      lcl      ucl event
#> 1          10 Treatment 1 6.495175 0.05711322 0.2389837 6.026776 6.963575   127
```

--------------------------------

### randomize_by_fixed_block examples with dplyr

Source: https://merck.github.io/simtrial/reference/randomize_by_fixed_block.html

Two examples in one block: first, a 2:1 randomization with block size 3 and treatments "A" and "B" applied to a data frame of 10 rows; second, stratified randomization by a stratum column with two groups, using the default block c(0,0,1,1). The output shows the resulting treatment assignments.

```r
library(dplyr)

# Example 1
# 2:1 randomization with block size 3, treatments "A" and "B"
data.frame(x = 1:10) |> mutate(Treatment = randomize_by_fixed_block(block = c("A", "B", "B")))
#>     x Treatment
#> 1   1         A
#> 2   2         B
#> 3   3         B
#> 4   4         B
#> 5   5         A
#> 6   6         B
#> 7   7         B
#> 8   8         B
#> 9   9         A
#> 10 10         A

# Example 2
# Stratified randomization
data.frame(stratum = c(rep("A", 10), rep("B", 10))) |>
  group_by(stratum) |>
  mutate(Treatment = randomize_by_fixed_block())
#> # A tibble: 20 × 2
#> # Groups:   stratum [2]
#>    stratum Treatment
#>    <chr>       <dbl>
#>  1 A               1
#>  2 A               1
#>  3 A               0
#>  4 A               0
#>  5 A               0
#>  6 A               0
#>  7 A               1
#>  8 A               1
#>  9 A               0
#> 10 A               0
#> 11 B               1
#> 12 B               0
#> 13 B               0
#> 14 B               1
#> 15 B               1
#> 16 B               1
#> 17 B               0
#> 18 B               0
#> 19 B               0
#> 20 B               0
```

--------------------------------

### Set up multisession backend with 2 workers

Source: https://merck.github.io/simtrial/articles/parallel.html

Initializes the multisession backend for parallel processing with two workers. By default, all available cores are used, but here we specify two.

```r
plan(multisession, workers = 2)
```

--------------------------------

### WLR test with FH weights and return_variance=TRUE

Source: https://merck.github.io/simtrial/reference/wlr.html

Performs the same WLR test as Example 1A but with return_variance=TRUE, which returns the variance-covariance matrix in addition to the standard output.

```r
x |> wlr(weight = fh(rho = 0, gamma = 0.5), return_variance = TRUE)
```

--------------------------------

### Example: Cut simulated data at 100 events and tabulate

Source: https://merck.github.io/simtrial/reference/cut_data_by_event.html

Simulates 200 observations with `sim_pw_surv()` and cuts the data at 100 events using the pipe operator. The resulting table shows the counts of censored (event = 0) and event (event = 1) subjects per treatment, with 50 events in each arm.

```R
# Use default enrollment and event rates at cut at 100 events
x <- sim_pw_surv(n = 200) |> cut_data_by_event(100)
table(x$event, x$treatment)
#>    
#>     control experimental
#>   0      48           49
#>   1      50           50

```

--------------------------------

### Simulate and cut survival data with sim_pw_surv and cut_data_by_event

Source: https://merck.github.io/simtrial/reference/wlr.html

Generates a simulated dataset with 200 patients and cuts it at the time of the 100th event. This dataset is used in the subsequent WLR test examples.

```r
x <- sim_pw_surv(n = 200) |> cut_data_by_event(100)
```

--------------------------------

### Run sim_fixed_n() with all timing types

Source: https://merck.github.io/simtrial/articles/routines.html

Simulates 2 trials with a sample size of 500 and target events of 350, using all five data cutoff methods. The output is formatted with gt() and fmt_number(). The backend uses sequential processing.

```R
sim_fixed_n(
  n_sim = 2, # Number of simulations
  sample_size = 500, # Trial sample size
  target_event = 350, # Targeted events at analysis
  stratum = stratum, # Study stratum
  enroll_rate = enroll_rate, # Enrollment rates
  fail_rate = fail_rate, # Failure rates
  total_duration = 30, # Planned trial duration
  block = block, # Block for treatment
  timing_type = 1:5, # Use all possible data cutoff methods
  rho_gamma = rho_gamma # FH test(s) to use; in this case, logrank
) |>
  gt() |>
  fmt_number(columns = c("ln_hr", "z", "duration"))
#> Backend uses sequential processing.
```

--------------------------------

### Set up input variables for sim_pw_surv()

Source: https://merck.github.io/simtrial/articles/routines.html

Defines stratum, block, enroll_rate, fail_rate, and dropout_rate for a trial with 400 patients. The strand division into Negative and Positive, block randomization with control/experimental arms, and piecewise constant enrollment and failure rates are configured here.

```R
stratum <- data.frame(stratum = c("Negative", "Positive"), p = c(.5, .5))

block <- c(rep("control", 2), rep("experimental", 2))

enroll_rate <- data.frame(rate = c(3, 6, 9), duration = c(3, 2, 1))

fail_rate <- data.frame(
  stratum = c(rep("Negative", 4), rep("Positive", 4)),
  period = rep(1:2, 4),
  treatment = rep(c(rep("control", 2), rep("experimental", 2)), 2),
  duration = rep(c(3, 1), 4),
  rate = log(2) / c(4, 9, 4.5, 10, 4, 9, 8, 18)
)
dropout_rate <- data.frame(
  stratum = c(rep("Negative", 4), rep("Positive", 4)),
  period = rep(1:2, 4),
  treatment = rep(c(rep("control", 2), rep("experimental", 2)), 2),
  duration = rep(c(3, 1), 4),
  rate = rep(c(.001, .001), 4)
)
```

--------------------------------

### rmst() with string column names

Source: https://merck.github.io/simtrial/reference/rmst.html

Example using the ex1_delayed_effect dataset with string column names. The output shows the RMST method, tau parameter, estimate, standard error, and z-statistic.

```r
data(ex1_delayed_effect)
rmst(
  data = ex1_delayed_effect,
  var_label_tte = "month",
  var_label_event = "evntd",
  var_label_group = "trt",
  tau = 10,
  reference = "0"
)
#> $method
#> [1] "RMST"
#> 
#> $parameter
#> [1] 10
#> 
#> $estimate
#> [1] 0.8650493
#> 
#> $se
#> [1] 0.3900344
#> 
#> $z
#> [1] 2.21788
#> 
```

--------------------------------

### Estimate single-arm RMST with simtrial:::rmst_single_arm

Source: https://merck.github.io/simtrial/articles/rmst.html

Estimates RMST for a single arm using the delayed effect example dataset, with tau set to 10, and displays the result as a gt table.

```r
data(ex1_delayed_effect)
data_single_arm <- ex1_delayed_effect[ex1_delayed_effect$trt == 1, ]
simtrial:::rmst_single_arm(
  time_var = data_single_arm$month,
  event_var = data_single_arm$evntd,
  tau = 10
) |> gt()
```

--------------------------------

### WLR with FH(0,1) using sim_fixed_n

Source: https://merck.github.io/simtrial/reference/sim_fixed_n.html

Runs a single simulation with timing_type=1 and rho=0, gamma=1. The output shows the WLR FH(rho=0, gamma=1) method with estimate, se, z, event, ln_hr, cut, duration, and sim columns. The power approximation is also shown as 0.

```R
# Example 2: WLR with FH(0,1) ----
sim_fixed_n(n_sim = 1, timing_type = 1, rho_gamma = data.frame(rho = 0, gamma = 1))
#> Backend uses sequential processing.
#>   method          parameter  estimate      se        z event      ln_hr
#> 1    WLR FH(rho=0, gamma=1) -1.608741 1.14817 1.401136    90 -0.2204257
#>                cut duration sim
#> 1 Planned duration       30   1
# Get power approximation
mean(x$z <= qnorm(.025))
#> [1] 0
```

--------------------------------

### Set up input parameters for sim_fixed_n()

Source: https://merck.github.io/simtrial/articles/routines.html

Defines stratum, enrollment rates, failure rates, treatment block, and rho_gamma for a 2-arm trial simulation. These parameters are used as arguments to sim_fixed_n().

```R
stratum <- data.frame(stratum = "All", p = 1)
enroll_rate <- data.frame(
  duration = c(2, 2, 10),
  rate = c(3, 6, 9)
)
fail_rate <- data.frame(
  stratum = "All",
  duration = c(3, 100),
  fail_rate = log(2) / c(9, 18),
  hr = c(0.9, 0.6),
  dropout_rate = rep(0.001, 2)
)
block <- rep(c("experimental", "control"), 2)
rho_gamma <- data.frame(rho = 0, gamma = 0)
```

--------------------------------

### cut_data_by_date function signature

Source: https://merck.github.io/simtrial/reference/cut_data_by_date.html

Usage signature for the cut_data_by_date function. It takes a time-to-event dataset x and a cut_date relative to the start of randomization (cte) and returns a data frame ready for survival analysis.

```r
cut_data_by_date(x, cut_date)
```

--------------------------------

### Get analysis date at 24 months with get_analysis_date

Source: https://merck.github.io/simtrial/reference/get_analysis_date.html

Calls get_analysis_date with planned_calendar_time = 24 and leaves all other arguments at their default NA. The function returns the analysis date as 24, matching the planned calendar time.

```R
get_analysis_date(
  simulated_data,
  planned_calendar_time = 24
)
#> [1] 24
```

--------------------------------

### Piecewise uniform enrollment simulation with rpwexp_enroll()

Source: https://merck.github.io/simtrial/reference/rpwexp_enroll.html

Simulates piecewise uniform enrollment for 100,000 patients with rates 5, 15, and 30 over durations 100, 200, and 100, then plots the enrollment times. The comment notes this approximates piecewise exponential inter-arrival times.

```r
# Example 1
# Piecewise uniform (piecewise exponential inter-arrival times) for 10k patients enrollment
# Enrollment rates of 5 for time 0-100, 15 for 100-300, and 30 thereafter
x <- rpwexp_enroll(
  n = 1e5,
  enroll_rate = data.frame(
    rate = c(5, 15, 30),
    duration = c(100, 200, 100)
  )
)
plot(x, 1:1e5,
  main = "Piecewise uniform enrollment simulation",
  xlab = "Time",
  ylab = "Enrollment"
)
```
Tokens:
Raw