Online Program Home
My Program

Abstract Details

Activity Number: 551 - Risk Prediction Methods and Applications in Risk Stratified Prevention
Type: Invited
Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
Sponsor: ENAR
Abstract #300050
Title: Using Deep Learning to Build Risk Prediction Models for Time-to-Event Outcomes
Author(s): Jon Steingrimsson* and Samantha Morrison and Constantine Gatsonis
Companies: Brown University and Brown University and Brown University
Keywords: Deep learning; Survival Analysis; Risk Prediction; Semi-parametric theory; Machine learning

Deep learning is a class of algorithms that uses multiple layers to create a risk prediction model. The layers involve an unknown weight vector that is estimated by minimizing a loss function. We extend the deep learning algorithms to handle censoring by replacing the loss function used in the absence of censoring by censoring unbiased loss functions. We discuss properties of these loss functions and practical issues related to implementation of the algorithms. The performance of the resulting deep learning algorithms is evaluated through simulation studies and analyzing data on breast cancer patients.

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

Back to the full JSM 2019 program