TrialSimulator

⚕ No for loop in clinical trial simulation ⚕

TrialSimulator is a system for declaratively implementing clinical trial simulations, inspired by the modular grammar of trial design. You provide the components—arms, endpoint-generation rules, milestones, and analysis methods—specify how they are combined under fixed or adaptive designs, and TrialSimulator takes care of the details, allowing you to focus on the design while it handles the mechanics.

TrialSimulator provides a suite of helper functions to support the full trial life cycle.

• For data generation, it offers utilities to sample endpoints from common distributions.
• For adaptation, it includes execution functions that modify trial settings in place.
• For analysis, it implements methods such as regression models, group sequential tests, graphical multiple testing procedures, combination tests, and closed tests. These tools are entirely optional, but they make it simple to construct and evaluate complex trial designs without sacrificing flexibility.

Shiny App

An online version of this app, TrialDesigner, is hosted on shinyapps.io. This app is still under development.

Talks

• Simulating Modern Clinical Trial in R: A Versatile Framework for Complex and Adaptive Trial Designs. Short course, 2026 NIC-ASA & ICSA Midwest Chapter Joint Fall Meeting, Northbrook, IL, October 15, 2026. Slides

Installation

Release

You can install the current release version from CRAN with:

install.packages("TrialSimulator")

Development

You can install the development version from GitHub with:

if(!requireNamespace("remotes", quietly = TRUE)){
  install.packages("remotes")
}
remotes::install_github(
  "zhangh12/TrialSimulator", 
  dependencies = TRUE, 
  upgrade = "never"
)

Getting Started

TrialSimulator is designed with a modular architecture, allowing users to systematically construct and simulate clinical trials in a structured and flexible manner. The simulation process is broken down into three key steps:

1. Defining Endpoints: Specify the treatment endpoints that will be evaluated throughout the trial.
2. Specifying Milestone: Establish the conditions under which specific milestones (e.g., interim analysis, futility assessment, or stopping rules) will be triggered.
3. Defining Actions for Milestones: Determine the appropriate actions to be taken once a milestone is triggered, such as adjusting enrollment, terminating treatment arms, or conducting statistical analysis.

To efficiently utilize TrialSimulator for clinical trial simulations, we recommend reading the following vignettes in order:

• Define endpoints and arms
  • Time-to-event (TTE) endpoints
    • Correlated PFS and OS using a Gumbel copula
    • Correlated PFS and OS under an illness-death model
  • Non-TTE endpoints
    • Special case: longitudinal endpoints
  • Define and summarize arms
• Condition system for milestones
  • Triggering trial milestones
• Define actions for milestones
  • Action function
• Examples: fixed design
  • Design with correlated endpoints, and custom data generator
• Examples: adaptive design
  • Seamless design with dose selection, interim, and multiple endpoints
  • Response-adaptive design
  • Dose-ranging study
  • Enrichment design with SSR
  • [Platform trial] vignette is under development.
  • [Basket trial] vignette is under development.
• Examples: crossover
  • Crossover trial with wash-out periods
  • Dynamic treatment switching
  • Crossover at milestone
• Built-in methods supported in TrialSimulator
  • Wrapper functions of common statistical methods for estimating treatment effect
  • [Group sequential test] See ?GroupSequentialTest.
  • [Graphical testing procedure] See ?GraphicalTesting.
  • [Combination test based on independent increment] extended from Jorgens et al. 2019. Vignette is under development.