Â鶹´«Ã½Ó³»­

Â鶹´«Ã½Ó³»­

Explore the latest content from across our publications
Join The Â鶹´«Ã½Ó³»­

Log In

Forgot Password?
Create New Account

Loading... please wait

Abstract Details

Network Analysis of Verbal Fluency in Individuals with Incident Cognitive Impairment
Aging, Dementia, and Behavioral Neurology
P16 - Poster Session 16 (5:30 PM-6:30 PM)
10-005
To analyze predictors of verbal fluency raw score and predictors of incident cognitive impairment (ICI) in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study dataset.
Verbal fluency is a brief cognitive test amenable to telephone administration and was collected in the REGARDS study on 25,625 participants over the age of 55. In addition, participants underwent annual procedures to evaluate general cognition, stroke risk, and incident stroke. Previous research suggests that contents of word lists generated during verbal fluency tasks have value for dementia prognosis. The REGARDS dataset includes verbal fluency for animals and F-words.

We identified 701 individuals without stroke but with incident cognitive impairment (ICI) based on performance on the Six-Item Screener who undertook both fluency tasks prior to ICI. Each subject in this ICI group was matched with a cognitively normal participant according to age, gender, region, and educational level. Following Goñi (2011), we derived a network defining the group-level word associations of each fluency task, then defined features based on graph traversal and quantified them for each individual word list. These features were then used as predictors in linear mixed effect models with raw score as the dependent variable and in Cox proportional hazards (CPH) models with ICI as the dependent variable.
All seven network-based features made significant contributions to variance in raw score for both letter F and animal fluency. In the CPH analyses, the only significant predictor of time to ICI was the length of the maximum inter-module chain in the animal fluency model. In the letter fluency analysis, the length of the maximum within-module chain showed a trend (p=0.075).
Network-based scores account for much of the variance in verbal fluency raw scores. Some network-based scores may serve to predict future cognitive impairment, but heterogeneity in the outcome may complicate the analysis.
Authors/Disclosures

PRESENTER 		No disclosure on file
No disclosure on file
Matthew R. Ayers, DO (Veterans Affairs) No disclosure on file
No disclosure on file
No disclosure on file
David G. Clark, MD Dr. Clark has received publishing royalties from a publication relating to health care.