Platform Trial Design II
It is known that sometimes at the design stage of a clinical trial, simulations are conducted to better understand the operating characteristics of the design. In a recently published paper, a simulation study was conducted for a platform trial design in the context of NASH (Non-Alcoholic Steatohepatitis). Different assumptions regarding e.g. treatment effect and correlation between the surrogate interim endpoint and the final endpoint, as well as design choices such as sample size and maximum number of cohorts allowed in the platform were investigated.
A description of the challenge can be found here.
A recording of the session can be found here.
Example 1. Shiny App - Gareth
Example 2. Exploratory visualisation
The app can be found here
Author: Steve Mallett
Example 3. Explanatory visualisation
Code
Example 1. Shiny App - Gareth
No code has been submitted.
Example 2. Exploratory visualisation
library(shiny)
library(tidyverse)
library(ggplot2)
library(ggbeeswarm)
library(introdataviz)
library(ggforce)
max_c <- c("All", "3", "5", "7")
cir <- c("All", "0.01", "0.03")
seq <- c("All", "0.065", "0.9", "1")
yn <- c("No", "Yes")
varlist1 <- c("Avg_Pat", "Avg_Pat_First_Suc", "Avg_Perc_Pat_Sup_Plac", "Avg_Cohorts", "FDR", "PTP", "PTT1ER","FWER", "FWER_BA", "Disj_Power", "Disj_Power_BA")
varlist2 <- c("ShortTermEndpointQuality", "Maximumnumberofcohorts", "TypeofDataSharing", "CohortInclusionRate", "FinalCohortSampleSize", "InterimFutilityStopping", "TreatmentEfficacySetting")
varlist3 <- c("None", "ShortTermEndpointQuality", "Maximumnumberofcohorts", "TypeofDataSharing", "CohortInclusionRate", "FinalCohortSampleSize", "InterimFutilityStopping", "TreatmentEfficacySetting")
varlist4 <- c("None", "CohortInclusionRate", "InterimFutilityStopping")
my_lgrey <-"#f0f0f0"
my_dgrey <- "#636363"
my_green <- "#639CA4FF"
my_orange <- "#BE7245FF"
fill_cols <- c(my_green, my_orange)
my_theme <- theme(text=element_text(
colour = my_dgrey,
size = 20),
panel.background=element_rect(fill="white"),
panel.border=element_rect(fill=NA),
panel.grid=element_line(colour = my_lgrey,
linewidth = 0.5,
linetype = 1),
axis.line=element_line(colour = my_dgrey,
linewidth = 0.5,
linetype = 1),
axis.text=element_text(
colour = my_dgrey,
size = 20),
axis.title=element_blank(),
plot.caption = element_text(hjust = 0, face= "italic"),
legend.title = element_blank()
)
ui <- fluidPage(
titlePanel("Platform Trial Design Tool"),
fluidRow(
column(2,
selectInput("yvar", "Y-Axis Variable", varlist1, selected="Disj_Power"),
selectInput("xvar", "X-Axis Variable", varlist2, selected="FinalCohortSampleSize"),
selectInput("trelvar", "Trellis Variable", varlist3),
selectInput("groupvar", "Group Variable", varlist4),
radioButtons("lg", "Log Transform?", yn),
),
column(5,
plotOutput("distPlot", width = "1200px", height = "1000px"))
))
server <- function(input, output) {
mydata <- read_csv("./data/ExampleDataNASH.csv")
output$distPlot <- renderPlot({
mydata <- mydata %>%
mutate(
yv = get(input$yvar),
xv = factor(get(input$xvar)))
if(input$trelvar != "None") {
mydata <- mydata %>%
mutate(
rv = factor(get(input$trelvar)))
}
if(input$groupvar != "None") {
mydata <- mydata %>%
mutate(
gv = factor(get(input$groupvar)))
}
if(input$groupvar == "None") {
my_plot <- ggplot() +
geom_violin(data=mydata, aes(x=xv, y=yv, group = xv),
fill=my_green, alpha=0.6, trim=FALSE)
}
else{
my_plot <- ggplot(mydata, aes(x=xv, y=yv,
group = interaction(gv, xv),
fill = gv)) +
geom_split_violin(alpha=0.6, trim=FALSE) +
scale_fill_manual(values=fill_cols)
}
if(input$trelvar != "None") {
my_plot <- my_plot +
facet_wrap(~rv)
}
if(input$lg == "Yes") {
my_plot <- my_plot +
scale_y_log10()
}
else {}
my_plot <- my_plot + my_theme
my_plot
})
}
# Run the application
shinyApp(ui = ui, server = server)Example 3. Explanatory visualisation
library(shiny)
library(tidyverse)
library(ggplot2)
library(introdataviz)
library(stringr)
library(ggforce)
my_lgrey <-"#f0f0f0"
my_dgrey <- "#636363"
my_green <- "#639CA4FF"
my_orange <- "#BE7245FF"
fill_cols <- c(my_green, my_orange)
my_theme <- theme(text=element_text(
colour = my_dgrey,
size = 14),
panel.background=element_rect(fill="white"),
panel.border=element_rect(fill=NA),
panel.grid=element_line(colour = my_lgrey,
linewidth = 0.5,
linetype = 1),
axis.line=element_line(colour = my_dgrey,
linewidth = 0.5,
linetype = 1),
axis.text=element_text(
colour = my_dgrey),
axis.title=element_text(colour = my_dgrey),
plot.caption = element_text(hjust = 0, face= "italic"),
legend.title = element_text(colour = my_dgrey,
size = 12),
plot.title = element_text(colour = my_dgrey)
)
# my_title <- "When the therapy is effective, inclusion of interim futility stopping rules reduces the disjunctive power when the short term endpoint quality is poor.\n"
my_cap <- "Treatment effect scenario: Both monotherapies are superior to SoC and combination therapy is better than monotherapies."
mydata <- read_csv("./data/ExampleDataNASH.csv") %>%
mutate(tes = if_else(TreatmentEfficacySetting==6, "Monotherapies > SoC. Combination > monotherapies", "Null Hypothesis")) %>%
mutate(steq = paste0("Short Term Endpoint Quality: ", ShortTermEndpointQuality)) %>%
filter(TreatmentEfficacySetting %in% c(1)) %>%
filter(ShortTermEndpointQuality %in% c(0.65, 1)) %>%
filter(FinalCohortSampleSize %in% c(100, 300, 500))
my_plot <- ggplot(mydata, aes(x=factor(FinalCohortSampleSize), y=Disj_Power,
group = interaction(InterimFutilityStopping, factor(FinalCohortSampleSize)),
fill = InterimFutilityStopping)) +
geom_sina(alpha=0.6, size=3, shape=21) +
# scale_fill_manual(values=fill_cols) +
scale_fill_manual("Interim\nFutility\nStopping?\n", values=fill_cols) +
scale_y_continuous("Disjunctive Power\n") +
scale_x_discrete("Final Cohort Sample Size\n") +
facet_wrap(~steq) +
labs(
# title = str_wrap(my_title, 200),
caption = str_wrap(my_cap, 100)) +
my_theme
ggsave("./WW_Jul24_03.png", my_plot, width=1250, height=900, units="px", dpi=150)