Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 149 - Statistical Learning for Decision Support
Type: Contributed
Date/Time: Monday, August 8, 2022 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #323020
Title: A Scalable and Flexible Cox Proportional Hazard Model for High-Dimensional Survival Prediction and Functional Selection
Author(s): Boyi Guo* and Nengjun Yi
Companies: University of Alabama at Birmingham and University of Alabama at Birmingham
Keywords: Cox Model; Spike-and-Slab; Scalable; R Package; Machine Learning; Interpretable
Abstract:

Cox proportional hazards (PH) model is one of the most popular models in biomedical data analysis. There have been continuing efforts to improve the flexibility of such models for complex signal detection, for example, via spline functions. Nevertheless, it is nontrivial to extend to the high-dimensional setting (p>>n). When estimating spline functions, commonly used grouped sparse regularization may induce excess shrinkage, damaging the predictive performance. Moreover, the previous “all-in-all-out” strategy for functional selection fails to answer if nonlinear components exist. We develop an additive Cox PH model that employs a novel spike-and-slab LASSO prior to select the linear and nonlinear components of spline functions. A scalable and deterministic algorithm, EM-Coordinate Descent, is designed for efficient model fitting. We compare the predictive and computational performance against the state-of-the-art models via Monte Carlo studies and metabolomics data analysis. The proposed model is broadly applicable to various research fields, e.g., genomics and population health, via the freely available R package BHAM (https://boyiguo1.github.io/BHAM/).


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program