|
Activity Number:
|
306
- Disease Prediction
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2018 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #330216
|
|
|
Title:
|
Grouping Trajectories of Unbalanced Longitudinal Data: a Comparison Between Growth Curve Mixture Models and Clustering BLUPs from Mixed Effects Models
|
|
Author(s):
|
Md Hossain* and Benjamin Leiby
|
|
Companies:
|
Nemours Biomedical Research, A.I. DuPont Children's Hospital and Thomas Jefferson University
|
|
Keywords:
|
Classification;
Unbalanced ;
longitudinal;
Growth Mixture Models;
BLUP
|
|
Abstract:
|
Clustering longitudinal data is a topic of growing interest in this data-driven decision-making era. While naturally occurring data such as electronic medical records have shown a great utility in making personalized medical decisions, methods available in major software packages are mostly applicable to data collected on a common set of occasions. Exploratory Growth Curve Mixture Models (GCMMs) have been used to identify subgroups of homogeneous trajectories in longitudinal unbalanced data. The method is unavailable in most standard statistical packages, requires specification of the number of classes prior to analysis, and may become computationally complex fairly quickly. A recent approach that applies conventional cluster methods to empirical best linear unbiased predictor of mixed linear spline models has been shown to have the ability to identify classes of homogeneous trajectories. The method resolves shortcomings of GCMMs. This study uses real and simulated datasets consisting of simple linear trends to complex temporal curves of irregular rates to compare classification performance of these two methods and suggest situations where one method outweigh the other.
|
Authors who are presenting talks have a * after their name.
