% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@INPROCEEDINGS{Amunts:863674,
author = {Amunts, Julia and Camilleri, Julia and Eickhoff, Simon and
Patil, Kaustubh and Heim, Stefan and Weis, Susanne},
title = {{P}redicting verbal fluency performance from executive
functions tests: {A} study in healthy subjects},
reportid = {FZJ-2019-03680},
year = {2019},
abstract = {Verbal fluency (VF) tasks are a common diagnostic tool in
neuropsychological assessment for testing executive
functions (EFs) and are part of several diagnostic batteries
in the clinical context. Interindividual differences like
age or sex impact performance of EFs leading to ambiguous
results concerning the relationship between EFs and VF.
Thus, the question of which subdomains of EFs impact on VF
tasks remains inconclusive. The present study examines
whether VF can be predicted on the basis of EF test
performance and aims to find out which variables have most
impact on prediction analysis.We tested 235 monolingual
German speaking healthy subjects (94 males, aged 20-55) with
13 EF tests and 3 VF tasks. These VF tests included two
simple and one switching task. Each task was performed for
two minutes and the total sum of corrects words was
calculated. Additionally, saliva samples were collected to
analyze for sex hormone levels and stress. Data were
adjusted for sex and age by linear regression and analyses
were computed on the residuals. For prediction analysis a
Relevance Vector Machine (RVM) approach with 500
replications and a 10-fold cross-validation was computed.
Prediction performance was assessed by computing the
correlation between real and predicted values. Variables
that had a non-zero weight in at least $80\%$ of all models
are considered to have the most influenced prediction
performance.A significant correlation was observed between
true and predicted values (r = 0.25; p < 0.001) and 9 EF
tests exhibited non-zero weights in more than $80\%$ of the
models. Out of these, the Stroop test, WAF-G (divided
attention), Raven´s Matrices test, Corsi and Wisconsin Card
Sorting test had at least $95\%$ non-zero weights. Stroop
contributed 3 features and WAF-G 2 out of the 9 relevant EFs
features. With regards to non-EFs features, Cortisol and
Estradiol had a high impact on prediction performance
$(100\%).Results$ of this study indicate that scores
obtained from tasks testing inhibition and updating are
representative to reflect VF performance. These findings
concur with previous studies which infer that updating
ability might be involved in VF when keeping track of
already spoken words to avoid errors and repetitions, while
inhibition might be needed to suppress repeated or
irrelevant responses. Additionally, together with other
studies, this study elucidates the impact of hormones on EFs
but also on VF performance. Previous studies have shown that
the menstrual cycle phase impacts accuracy and processing
speed in EFs test while the effects of cortisol are still
inconclusive.},
month = {Jun},
date = {2019-06-25},
organization = {INM/ICS retreat 2019, Jülich
(Germany), 25 Jun 2019 - 26 Jun 2019},
subtyp = {After Call},
cin = {INM-7 / INM-1},
cid = {I:(DE-Juel1)INM-7-20090406 / I:(DE-Juel1)INM-1-20090406},
pnm = {571 - Connectivity and Activity (POF3-571) / 572 -
(Dys-)function and Plasticity (POF3-572) / 574 - Theory,
modelling and simulation (POF3-574)},
pid = {G:(DE-HGF)POF3-571 / G:(DE-HGF)POF3-572 /
G:(DE-HGF)POF3-574},
typ = {PUB:(DE-HGF)24},
url = {https://juser.fz-juelich.de/record/863674},
}