Survival-Pipe: Diagnostic-First Survival Analysis

A Modular Pipeline for Time-to-Event Data

Author

Hsieh-Ting Lin

Published

March 28, 2026

Welcome

Survival-Pipe is a modular R + Python pipeline for survival analysis in clinical research. It takes a diagnostic-first approach: before fitting any model, the pipeline inspects your data for competing risks, recurrent events, time-varying exposures, left truncation, informative censoring, and clustering — then routes you to the correct analysis automatically.

Key Features

  • Diagnostic-first: Six automated checks run before any model is fit, preventing misspecified analyses.
  • Modular: Ten sa-* modules cover the full workflow from data intake to manuscript rendering.
  • Reproducible: Every step is scripted, versioned, and orchestrated via make targets.
  • Publication-ready: 300 DPI figures, AMA-styled manuscripts, and Quarto-rendered PDF/DOCX output.

Decision Flowchart

flowchart TD
    A[Start: Raw Data] --> B[sa-data-intake<br/>Clean + Validate]
    B --> C[sa-diagnostics<br/>6 Automated Checks]
    C --> D{Competing<br/>risks?}
    D -- Yes --> E[sa-competing-risks]
    D -- No --> F{Recurrent<br/>events?}
    F -- Yes --> G[sa-recurrent-multistate]
    F -- No --> H{Time-varying<br/>exposure?}
    H -- Yes --> I[sa-time-varying]
    H -- No --> J[sa-standard-km]
    E --> K{Adjustments<br/>needed?}
    G --> K
    I --> K
    J --> K
    K -- Yes --> L[sa-advanced-adjustments]
    K -- No --> M[sa-publication-figures]
    L --> M
    M --> N[sa-manuscript-quarto<br/>Render Paper]