Activity Number:
|
653
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 4, 2016 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #321091
|
View Presentation
|
Title:
|
A Hidden Markov Model Approach to Analyzing Longitudinal Alzheimer's Disease Stages Subject to Possible Misclassification
|
Author(s):
|
Julia Benoit* and Wenyaw Chan and Linda Piller and Rachelle S. Doody
|
Companies:
|
University of Houston and The University of Texas Health Science Center at Houston and The University of Texas Health Science Center at Houston and Baylor College of Medicine
|
Keywords:
|
continuous-time markov chain ;
EM algorithm ;
misclassification
|
Abstract:
|
Understanding the dynamics of a disease process is essential in early detection, diagnosis, and measuring progression. Continuous-time Markov chain (CTMC) methods have been used to estimate state change rates (or intensities) in various fields but challenges arise when stages are potentially misclassified. A likelihood approach we recently developed is presented here where the target outcome is modeled as hidden states of a 3-state CTMC model using the possibly misclassified observed values. Covariate effects of the underlying stochastic process and probabilities of misclassification of the target state are estimated without information from a 'gold standard' as comparison. We present an application which estimates the probability of misclassification and finds the determinants of Alzheimer's disease stage changes over time using current classifications of AD which could be misclassified. This research provides a comparison of sensitivity of staging disease and aims to corroborate cut-points; and is the first as we know to integrate longitudinal data to estimate sensitivity of cognitive disease severity.
|
Authors who are presenting talks have a * after their name.