SurvControl

An interactive control panel for survival analysis on real oncology data.

🔗 Live demo  ·  R / Shiny

SurvControl is a dashboard that lets users explore time-to-event (survival) data without writing code. It fits and compares several survival models, visualizes how covariates affect survival, and tests whether censoring is dependent on the event of interest, a case where standard models can mislead.

Built as my undergraduate thesis in Statistics at UFRGS (advisor: Profa. Dra. Silvana Schneider, 2023).

What it does

• Explore covariate effects (age, sex, education, clinical stage, etc.) on survival across groups or individuals.
• Fit and compare models: Kaplan-Meier estimator, Cox regression, frailty models, and dependent-censoring models (Weibull and piecewise-exponential marginals, via the DepCens package).
• Diagnose dependent censoring — detect when failure time and censoring time are associated, and quantify why ignoring it biases results.
• Visualize survival and cumulative-hazard curves interactively.

Data

Applied to anonymized hospital cancer-registry data (RHC) from the Fundação Oncocentro de São Paulo (FOSP, 2022), with a comparative case study of breast (C50) and ovarian (C56) cancer patients. The analysis found a meaningful association between failure and censoring times — confirming the value of dependent-censoring models for this data.

Getting Started

Requirements: R ≥ 4.0

# 1. Install all dependencies (run once)
source("install.R")

# 2. Launch the app
shiny::runApp("app.R")

Step 1 only need to be run once after cloning. After that, only step 2 is needed.

Tech

R · Shiny · survival · DepCens