R Survival Mini‑Kit — Kaplan–Meier, Cox PH, LASSO‑Cox (RMarkdown)

A compact, portfolio‑ready R project that showcases clinical survival analysis the R way:

• Kaplan–Meier curves with log‑rank tests
• Cox Proportional Hazards (PH) model
• Regularized Cox (LASSO) with glmnet
• Risk stratification by linear predictor quantiles
• Time‑dependent ROC (adaptive horizons with fallback)
• Clean RMarkdown report for presentation

The goal is to demonstrate practical R skills in a domain where R is naturally strong (clinical/biostat). It's intentionally small and reproducible.

---

What’s inside

r-survival-mini-kit/
├─ R/
│  ├─ prep.R          # load/clean the dataset (lung/ovarian)
│  ├─ km_plots.R      # KM curves + log-rank
│  ├─ cox_models.R    # Cox PH + glmnet LASSO-Cox
│  └─ eval.R          # risk strata, timeROC + fallback to riskRegression
├─ report.Rmd         # main HTML report (render this)
├─ report.html        # rendered report (kept in repo for convenience)
├─ README.md
├─ .gitignore
├─ .Rprofile          # optional: auto-activate renv if present
└─ DESCRIPTION        # minimal project metadata

---

Quick start

1. Open the folder in RStudio (or any R environment).
2. Install packages (first run):

install.packages(c(
  "survival","glmnet","timeROC","broom","tidyverse","gt","rmarkdown","riskRegression"
))

3. Knit the report:

rmarkdown::render("report.Rmd", output_format = "html_document")

The output HTML will be saved alongside the Rmd.

Optional: use renv to lock dependencies later (install.packages("renv"); renv::init()), but it’s not required for the first run.

---

Reproducibility

This report is deterministic:

• We fix the global RNG seed at the start of the report:

  # in report.Rmd setup chunk
  set.seed(42)

• The report was rendered on R 4.3.x (see the Session info section in the HTML).
  Using a different R version may cause minor differences in CV splits/plots.

Recommended options (optional):

• Pin package versions via renv:

  install.packages("renv")
  renv::init()          # snapshot current package versions
  # later on another machine:
  renv::restore()       # reproduce the environment

• Capture exact session metadata in the report footer (already included):

  sessionInfo()

---

Why this project?

• R in its native habitat (clinical stats) — no “forcing R through the window”
• Crisp and short: suitable as a side project and a clean GitHub repo
• Clear structure with reusable functions (R/ folder) and a single presentable report

---

Suggested repo metadata

• Name: r-survival-mini-kit
• Description: Compact survival analysis in R: KM, log‑rank, Cox PH, LASSO‑Cox, time‑ROC. Clean RMarkdown report.
• Topics: r, survival-analysis, cox, glmnet, clinical-data, tidyverse, rmarkdown, portfolio-project

---

License

MIT — see LICENSE.