Smooth transformation models for survival analysis: A tutorial using R

Area of Science:

• Statistics
• Biostatistics
• Computational Statistics

Background:

• Survival analysis has advanced significantly over five decades.
• Navigating diverse methods and software for complex survival data (e.g., interval-censored, clustered, non-proportional hazards) is challenging.
• A unified framework is needed to manage evolving methodological landscapes.

Purpose of the Study:

• To explore the utility of smooth transformation models for survival analysis within the R statistical computing environment.
• To demonstrate a unified maximum-likelihood approach for various survival models and complex scenarios.
• To showcase the practical application of the `tram` package in R for navigating survival analysis tasks.

Main Methods:

• Utilized the framework of smooth transformation models for survival analysis.
• Employed a unified maximum-likelihood estimation approach.
• Applied the `tram` package and related R packages to analyze survival data from a rectal cancer clinical trial.

Main Results:

• The smooth transformation model framework in R provides a unified approach to various survival models, including Weibull and Cox proportional hazards models.
• This framework successfully accommodates complex scenarios such as non-proportional hazards, dependent censoring, and clustered data.
• The `tram` package facilitates seamless navigation of survival analysis tasks for real-world data, including personalized medicine extensions.

Conclusions:

• Smooth transformation models offer a versatile and unified framework for survival analysis in R.
• The `tram` package provides an efficient implementation for handling both standard and complex survival data structures.
• This approach simplifies the application of advanced survival methods, aiding researchers in complex data analysis and clinical trial interpretation.