|
Activity Number:
|
154
- Contributed Poster Presentations: Lifetime Data Science Section
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 8, 2022 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Lifetime Data Science Section
|
|
Abstract #322438
|
|
|
Title:
|
Comparing Individualized Survival Predictions from Random Survival Forests and Multistate Models: A Case Study of Patients with Oropharyngeal Cancer
|
|
Author(s):
|
Madeline Abbott* and Lauren Beesley and Jeremy Taylor
|
|
Companies:
|
University of Michigan and University of Michigan and University of Michigan
|
|
Keywords:
|
survival analysis;
multistate model;
random survival forest;
cancer application
|
|
Abstract:
|
In recent years, interest in prognostic prediction calculators has grown with the popularity of personalized medicine. These calculators, which can inform treatment decisions, employ many different methods, each of which has advantages and disadvantages. We present a comparison of a highly-structured parametric Bayesian multistate model (MSM) and a black-box non-parametric random survival forest (RSF) through a case study of prognostic predictions for patients with oropharyngeal squamous cell carcinoma. Key in this comparison is the high rate of missing values within these data and the different approaches used by the MSM and RSF to account for these missing values. We compare the accuracy (discrimination and calibration) of survival probabilities predicted by both approaches and use simulation studies to better understand how predictive accuracy is influenced by (1) the handling missing data and (2) the modeling of structural/disease progression information present in the data. We conclude that both approaches have similar predictive accuracy and that consideration of other differences (e.g. interpretability) are key when selecting the best approach.
|
Authors who are presenting talks have a * after their name.
