|
Activity Number:
|
190
- Synergy of Bayesian, Frequentist, and Fiducial Approaches in Addressing Modern Statistical Problems in Data Science
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 10, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
|
Abstract #316692
|
|
|
Title:
|
Robust and individualized conformal predictive inference in survival analysis
|
|
Author(s):
|
Minge Xie*
|
|
Companies:
|
Rutgers University
|
|
Keywords:
|
Conformal Prediction;
Survival Analysis;
Individualized Inference
|
|
Abstract:
|
This talk presents a novel predictive inference approach to provide prediction intervals for the survival time of a given patient in survival analysis. The development is based on a machine learning method known as conformal prediction. Compared with existing survival analysis approaches, the new development simultaneously enjoys two nice features: the prediction bounds provided are individualized (i.e., different new patients have their own predictions); and (2) the inference results remain valid even if the survival model used is wrong. In addition to the standard conformal prediction development in which the inference statement is on the joint probability of covariates and survival events, we also use the so-called individualized fusion (iFusion) learning method to develop the second type of prediction intervals in which the inference statement is conditional on individual covariates. The research is motivated by the need to support clinical research on Alzheimer disease through mining the OneFlorida network data. The utility of the proposed methodology is demonstrated through simulations and real data examples.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2021 program
|
