Estimating Penalized AFT Models via Coordinate Descent.

aftPenCDA

aftPenCDA is an R package for fitting penalized accelerated failure time (AFT) models using induced smoothing and coordinate descent algorithms. Computationally intensive components are implemented in 'C++' via 'Rcpp' (RcppArmadillo backend) to ensure scalability in high-dimensional settings.

The package supports both right-censored survival data and clustered partly interval-censored survival data, and provides flexible variable selection through several penalty functions.

Features

• Induced smoothing-based AFT estimation for nonsmooth estimating equations
• Penalized coordinate descent algorithm
• Efficient C++ implementation via Rcpp (RcppArmadillo)
• Support for high-dimensional regression
• Supported penalties:
  • Broken Adaptive Ridge (BAR)
  • LASSO
  • Adaptive LASSO (ALASSO)
  • SCAD
• Variance estimation methods:
  • "CF": closed-form plug-in estimator
  • "ZL": perturbation-based estimator based on Zeng and Lin (2008)

Installation

You can install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("seonsy/aftPenCDA")

Main functions

aftpen()

Fits a penalized AFT model for right-censored survival data.

aftpen_pic()

Fits a penalized AFT model for clustered partly interval-censored survival data.

Both functions share the same interface:

aftpen(dt, lambda = 0.1, se = "CF", type = "BAR")
aftpen_pic(dt, lambda = 0.1, se = "CF", type = "BAR")

Input data format

Right-censored data (aftpen())

• dt: a data frame where:
  • first column: y (observed time)
  • second column: d (event indicator; 1 = event, 0 = censoring)
  • remaining columns: covariates

Clustered partly interval-censored data (aftpen_pic())

• dt: a data frame containing:
  • L, R: interval endpoints
  • delta: exact observation indicator (1 = exact, 0 = censored)
  • id: cluster identifier
  • remaining columns: covariates

Algorithm

The method combines induced smoothing with a coordinate descent algorithm. A quadratic approximation is constructed via Cholesky decomposition, leading to a least-squares-type problem.

Efficient coordinate-wise updates are then applied under different penalties.

Example

Right-censored data

library(aftPenCDA)

data("simdat_rc")

fit <- aftpen(simdat_rc, lambda = 0.1, se = "CF", type = "BAR")

fit$beta

Clustered partly interval-censored data

data("simdat_pic")

fit_pic <- aftpen_pic(simdat_pic, lambda = 0.001, se = "CF", type = "BAR")

fit_pic$beta

Arguments

Value

Both functions return a list with components:

• beta: final penalized coefficient estimate

References

Wang, You-Gan, and Yudong Zhao. 2008. “Weighted Rank Regression for Clustered Data Analysis.” Biometrics 64 (1): 39–45.

Dai, L., K. Chen, Z. Sun, Z. Liu, and G. Li. 2018. “Broken Adaptive Ridge Regression and Its Asymptotic Properties.” Journal of Multivariate Analysis 168: 334–351.

Zeng, Donglin, and D. Y. Lin. 2008. “Efficient Resampling Methods for Nonsmooth Estimating Functions.” Biostatistics 9 (2): 355–363.

Note

This package is under development. Functionality and interfaces may change in future versions.