% 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”.
@ARTICLE{Gell:1052292,
author = {Gell, Martin and Hoffmann, Mauricio S. and Moore, Tyler M.
and Nikolaidis, Aki and Gur, Ruben C. and Salum, Giovanni A.
and Milham, Michael P. and Langner, Robert and Müller,
Veronika I. and Eickhoff, Simon B. and Satterthwaite,
Theodore D. and Tervo-Clemmens, Brenden},
title = {{D}isentangling {B}rain-{P}sychopathology {A}ssociations:
{A} {S}ystematic {E}valuation of {T}ransdiagnostic
{B}ifactor {M}odels},
reportid = {FZJ-2026-00909},
year = {2025},
abstract = {Understanding the neurobiological basis of mental health
disorders remains a central goal in psychiatry. However,
identifying robust brain-psychopathology associations with
neuroimaging has proven difficult, in part due to
substantial heterogeneity within and comorbidity between
diagnostic categories. Transdiagnostic bifactor models aim
to characterise this structure by separating shared from
unique symptom variance, yielding more reliable and
potentially more accurate latent dimensions of
psychopathology. However, the extent to which these
behavioural models improve brain-psychopathology
associations remains largely uncharacterised. Using two
large developmental cohorts, we compared 11 previously
published bifactor models applied to the Child Behaviour
Checklist (CBCL) to traditional CBCL summary scores. For
both symptom-scoring approaches, we systematically evaluated
their reliability and multivariate associations with
whole-brain structure (MRI) and function (resting-state
fMRI). We found no consistent evidence that bifactor-derived
factor scores strengthened reliability or
brain-psychopathology associations, relative to summary
scores. Whole-brain predictive models revealed broadly
distributed neural signatures that were highly similar
between corresponding factor and summary score constructs,
with general factors and total problems approaching
numerical equivalence. Nevertheless, factor scores displayed
more distinct neural signatures between general,
internalising, and externalising dimensions than did summary
scores. Together, these findings suggest that existing CBCL
bifactor models of psychopathology do not systematically
strengthen the predictive utility of psychiatric
neuroimaging, possibly reflecting fundamental limits on the
proportion of CBCL symptom variance captured by brain
features. While bifactor models may aid in separating neural
correlates across constructs, improving phenotypic
assessment depth, rather than alternative phenotypic
modelling, may provide more tangible improvements to
association strength moving forward.},
cin = {INM-7},
cid = {I:(DE-Juel1)INM-7-20090406},
pnm = {5251 - Multilevel Brain Organization and Variability
(POF4-525)},
pid = {G:(DE-HGF)POF4-5251},
typ = {PUB:(DE-HGF)25},
doi = {10.64898/2025.12.21.695029},
url = {https://juser.fz-juelich.de/record/1052292},
}