Activity Number:
|
257
- SPEED: Longitudinal/Correlated Data
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 2:00 PM to 2:45 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #332630
|
|
Title:
|
Comparisons of Modeling Methods on Longitudinal and Survival Data: Identifying Use of Repeat Biomarker Measurements to Predict Time-To-Event Outcome in Cancer Research
|
Author(s):
|
Meng Ru* and Erin Moshier and Madhu Mazumdar
|
Companies:
|
Icahn School of Medicine at Mount Sinai and Icahn School of Medicine at Mount Sinai and Icahn School of Medicine at Mount Sinai
|
Keywords:
|
time-to-event outcome;
bias adjustment;
longitudinal analysis;
latent class;
biomarker;
joint model of longitudinal and survival data
|
Abstract:
|
There has been growing interest in the use of statistical methods to associate longitudinal exposures with a time-to-event outcome. While the simplest approach is the time-dependent Cox model, it can produce biased risk estimates due to measurement error and sparsity of longitudinal data. To address these limitations more sophisticated methods including joint modeling (JM) and two-stage approaches have been developed. The JM approach models the longitudinal trend(s) and time-to-event data simultaneously in one-stage typically using a shared random effects model while the two-stage approach models the longitudinal process in stage 1 by a mixed effects or latent trajectory model and then separately fits the survival process in stage 2 by using a landmark Cox model with the individual random-effects or group assignments obtained from stage 1 as covariates. Appropriate methods for predicting time to event outcomes using longitudinal data are underutilized in the oncological sciences. We illustrate the performance of these approaches using data of repeat biomarkers collected from multiple myeloma patients to predict progression.
|
Authors who are presenting talks have a * after their name.