user2024-survival-mlr3

This introductory tutorial is designed to equip participants with practical skills and knowledge for performing survival analysis using machine learning techniques. Survival analysis, a fundamental statistical method in biomedical and clinical research, focuses on analyzing time-to-event data, such as the time to disease progression or patient survival. In this tutorial, attendees will work with clinical and gene expression data to build, train, and test survival models. They will learn how to leverage R’s {mlr3} ecosystem for efficient model development, incorporating sophisticated machine learning models such as penalized linear models and random forests to enhance the accuracy of the survival predictions. Participants will also explore survival metrics and model validation techniques to assess the quality and reliability of their models in the context of real-world data. Whether you’re new to survival analysis or seeking to enhance your skills, this workshop offers valuable insights and hands-on experience for tackling challenging clinical and biomedical questions.

Learning Goals

• Understand the foundations of Survival Analysis and its applications in clinical and high-dimensional research.
• Develop skills in using the {mlr3} framework for survival analysis, allowing you to build and evaluate predictive models.
• Explore the various survival prediction types and survival metrics to assess model performance.
• Work with real-world clinical and gene expression datasets to apply machine learning techniques in a research context.

Requirements for participating:

Bring a laptop with R 4.4.0 installed.
Install/update the following packages after 1 July to ensure you have the latest versions:

• tidyverse (CRAN)
• mlr3verse (CRAN) (incl. paradox >= 1.0.0)
• mlr3viz (GitHub mlr-org/mlr3viz, latest version for new features!)
• survival (CRAN)
• survminer (CRAN)
• rpart (CRAN)
• mlr3proba (GitHub mlr-org/mlr3proba)
• mlr3extralearners (GitHub mlr-org/mlr3extralearners)

Install packages for models that we will try:

• glmnet (CRAN) (Required)
• ranger (CRAN) (Optional)
• aorsf (CRAN) (Optional)
• CoxBoost (GitHub binderh/CoxBoost) (Optional)

Install all packages with {pak} (install.packages("pak")):

# CRAN packages first
pak::pak(c("tidyverse", "mlr3verse", "survival", "rpart", "glmnet", "ranger", "aorsf", "survminer", upgrade = TRUE)

# Non-CRAN packages from GitHub
pak::pak(c("mlr-org/mlr3proba", "mlr-org/mlr3extralearners", "mlr-org/mlr3viz"), upgrade = TRUE)
pak::pak("binderh/CoxBoost")

Please note that we will have limited time to help with package installation issues during the workshop.
We recommend installing the packages in advance to ensure a smooth experience.

References

1. Bischl, B., Sonabend, R., Kotthoff, L., & Lang, M. (Eds.). (2024). “Applied Machine Learning Using mlr3 in R”. CRC Press. https://mlr3book.mlr-org.com
2. Sonabend, R., Király, F. J., Bender, A., Bischl, B., & Lang, M. (2021). mlr3proba: an R package for machine learning in survival analysis. Bioinformatics, 37(17), 2789–2791. https://doi.org/10.1093/BIOINFORMATICS/BTAB039
3. Zhao, Z., Zobolas, J., Zucknick, M., & Aittokallio, T. (2024). Tutorial on survival modeling with applications to omics data. Bioinformatics. https://doi.org/10.1093/BIOINFORMATICS/BTAE132 Tutorial link