Predicting future risk of developing cognitive impairment using ambulatory sleep EEG: Integrating univariate analysis and multivariate information theory approach
Journal of Alzheimer s Disease,
Journal Year:
2025,
Volume and Issue:
unknown
Published: March 4, 2025
Background
Early
identification
of
individuals
at
risk
for
cognitive
impairment
is
crucial,
as
the
preclinical
phase
offers
an
opportunity
interventions
to
slow
disease
progression
and
improve
outcomes.
Objective
While
sleep
electroencephalography
(EEG)
has
shown
significant
promise
in
detecting
impairment,
this
study
aims
1)
develop
validate
overnight
EEG
biomarkers
prediction
future
risk,
2)
assess
their
predictive
performance
within
5
years,
3)
explore
feasibility
using
wearable,
low-density
devices
convenient
at-home
monitoring.
Methods
Overnight
polysomnography
was
performed
on
281
cognitively
normal
women
Study
Osteoporotic
Fractures
(SOF).
Cognitive
reassessments
were
conducted
approximately
five
years
later.
Features
such
relative
power
across
different
frequency
bands
channel
interactions,
quantified
generalized
mutual
information
measures,
extracted
used
inputs
machine
learning
models.
Binary
classification
models
distinguished
participants
who
developed
from
those
remained
normal.
Optimal
feature
subsets
classiffiation
identifed,
with
additional
analyses
testing
contribution
demographic
data,
macrostructure,
APOE
genotype.
Results
The
optimal
model,
utilizing
univariate
multivariate
features,
achieved
AUC
0.76.
N3
stage
gamma
band
exhibited
largest
effect
sizes.
Adding
demographics,
genotype
did
not
enhance
performance.
Conclusions
demonstrate
a
promising,
cost-effective
approach
early
assessment.
Larger
studies
more
diverse
populations
are
required
expand
these
findings
populations.
Language: Английский
Graph Laplacian Spectrum of Structural Brain Networks is Subject-Specific, Repeatable but Highly Dependent on Graph Construction Scheme
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2023,
Volume and Issue:
unknown
Published: June 4, 2023
Abstract
It
has
been
proposed
that
the
estimation
of
normalized
graph
Laplacian
over
a
brain
network’s
spectral
decomposition
can
reveal
connectome
harmonics
(eigenvectors)
corresponding
to
certain
frequencies
(eigenvalues).
Here,
I
used
test-retest
dMRI
data
from
Human
Connectome
Project
explore
repeatability,
and
influence
construction
schemes
on
a)
spectrum,
b)
topological
properties,
c)
high-order
interactions,
d)
their
associations
structural
network
(SBN).
Additionally,
investigated
performance
subject’s
identification
accuracy
(brain
fingerprinting)
interactions.
Normalized
eigenvalues
were
found
be
subject-specific
repeatable
across
five
schemes.
The
repeatability
is
lower
than
shows
heavy
dependency
scheme.
A
relationship
between
specific
properties
SBN
with
spectrum
was
also
revealed.
absolute
(100%)
schemes,
while
similar
observed
for
combination
(communities,3,4-motifs)
only
9m-OMST.
Collectively,
interactions
characterized
uniquely
SBN.
Language: Английский