• Project Due April 30 by email
• About 5 written pages + code + plots
  • Format could vary a bit depending on project e.g. if you implement reasonably complex model in R, final report could be 2 pages describing model + link to github repo with model code and testing functions
• Topic: Anything in the area of survival analysis
• Email me 1 paragraph description of project plan by April 15
  • Discuss report format
• The following slides contain some project ideas. You do not need to pursue one of these ideas.
  • Mix different project suggestions together.
  • Already have an idea? Go with it.

Acquire data set other than KIRC from the TCGA website and fit models similar to homework 5.

Suggested Report Format:

• 3 pages describing data acquisition and cleaning process
• 2 pages applying Homework 5 code to data and summarizing results

Fit models other than Cox PH to TCGA KIRC data (or another data set). Some examples:

• DeepSurv Deep learning for survival analysis outcomes.
• Survival Trees Regression Trees for Survival Data
  • party package
• Survival Forests Random forests for survival data.
• many others . . .

Suggested Report Format:

• 2 pages describing model
• 3 pages applying model to data, comparing with Cox PH

• Bayesian Accelerated Failure Time Model With Multivariate Doubly Interval-Censored Data and Flexible Distributional Assumptions: Summarize model and reproduce result in work.
• CRAN Survival Task View: Describe model and run simulation (e.g. comparing speed, estimation accuracy, performance under model misspecification) for package in "Bayesian Models"

• Summarize model and reproduce simulation results for Causal Inference on the Difference of the Restricted Mean Lifetime Between Two Groups
• Summarize model and reproduce results for Mediation Analysis for Censored Survival Data under an Accelerated Failure Time Model

• Generating survival times to simulate Cox proportional hazards models: Describe methods and run simulations for generating data from Cox PH model
• Survival analysis with high-dimensional covariates: Describe the methods in Table 3 and reproduce Figure 4.
• BAP1 links metabolic regulation of ferroptosis to tumour suppression: Reproduce pieces of Figure 2, especially survival curves
• Gene Expression Profiling Predicts Survival in Conventional Renal Cell Carcinoma: Obtain data set from GEO database and reproduce figures or fit regularized Cox model