Predicting
an
individual's
cognitive
traits
or
clinical
condition
using
brain
signals
is
a
central
goal
in
modern
neuroscience.
This
commonly
done
either
structural
aspects,
such
as
connectivity
cortical
thickness,
aggregated
measures
of
activity
that
average
over
time.
But
these
approaches
are
missing
aspect
function:
the
unique
ways
which
unfolds
One
reason
why
dynamic
patterns
not
usually
considered
they
have
to
be
described
by
complex,
high-dimensional
models;
and
it
unclear
how
best
use
models
for
prediction.
We
here
propose
approach
describes
functional
amplitude
Hidden
Markov
model
(HMM)
combines
with
Fisher
kernel,
can
used
predict
individual
traits.
The
kernel
constructed
from
HMM
mathematically
principled
manner,
thereby
preserving
structure
underlying
model.
show
here,
fMRI
data,
HMM-Fisher
accurate
reliable.
compare
other
prediction
methods,
both
time-varying
time-averaged
connectivity-based
models.
Our
leverages
information
about
has
broad
applications
neuroscience
personalised
medicine.
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2023,
Номер
unknown
Опубликована: Янв. 20, 2023
Abstract
Working
memory
(WM)
plays
a
central
role
in
cognition,
prompting
neuroscientists
to
investigate
its
functional
and
structural
substrates.
The
WM
dynamic
recruits
large-scale
frequency-specific
brain
networks
that
unfold
over
few
milliseconds
–
this
complexity
challenges
traditional
neuroimaging
analyses.
In
study,
we
unravel
the
network
dynamics
an
unsupervised,
data-driven
way,
applying
time
delay
embedded-hidden
Markov
model
(TDE-HMM).
We
acquired
MEG
data
from
38
healthy
subjects
performing
n-back
working
task.
TDE-HMM
inferred
four
task-specific
states
with
each
unique
temporal
(activation),
spectral
(phase-coherence
connections),
spatial
(power
density
distribution)
profiles.
A
theta
frontoparietal
state
performs
executive
functions,
alpha
temporo-occipital
maintains
information,
broad-band
spatially
complex
M300
profile
leads
retrieval
process
motor
response.
HMM
can
be
straightforwardly
interpreted
within
neuropsychological
multi-component
of
WM,
significantly
improving
comprehensive
description
WM.
Highlights
different
wax
wane
at
millisecond
scale.
Through
time-delay
embedded
hidden
(TDE-HMM)
are
able
extract
spatial,
spectral,
demonstrate
existence
well-known
Baddeley’s
multicomponent
memory.
This
novel
unveils
new
features
will
lead
more
in-depth
characterization
cognitive
processes
data.
Journal of Neural Engineering,
Год журнала:
2023,
Номер
20(4), С. 046007 - 046007
Опубликована: Июнь 20, 2023
Abstract
Objectives
.
Recent
event-based
analyses
of
transient
neural
activities
have
characterized
the
oscillatory
bursts
as
a
signature
that
bridges
dynamic
states
to
cognition
and
behaviors.
Following
this
insight,
our
study
aimed
(1)
compare
efficacy
common
burst
detection
algorithms
under
varying
signal-to-noise
ratios
event
durations
using
synthetic
signals
(2)
establish
strategic
guideline
for
selecting
optimal
algorithm
real
datasets
with
undefined
properties.
Approach.
We
tested
robustness
simulation
dataset
comprising
multiple
frequencies.
To
systematically
assess
their
performance,
we
used
metric
called
‘detection
confidence’,
quantifying
classification
accuracy
temporal
precision
in
balanced
manner.
Given
properties
empirical
data
are
often
unknown
advance,
then
proposed
selection
rule
identify
an
given
validated
its
application
on
local
field
potentials
basolateral
amygdala
recorded
from
male
mice
(n=8)
exposed
natural
threat.
Main
Results.
Our
simulation-based
evaluation
demonstrated
is
contingent
upon
duration,
whereas
accurately
pinpointing
onsets
more
susceptible
noise
level.
For
data,
chosen
based
exhibited
superior
accuracy,
although
statistical
significance
differed
across
frequency
bands.
Notably,
by
human
visual
screening
one
recommended
rule,
implying
potential
misalignment
between
priors
mathematical
assumptions
algorithms.
Significance.
Therefore,
findings
underscore
precise
fundamentally
influenced
algorithm.
The
algorithm-selection
suggests
potentially
viable
solution,
while
also
emphasizing
inherent
limitations
originating
algorithmic
design
volatile
performances
datasets.
Consequently,
cautions
against
relying
solely
heuristic-based
approaches,
advocating
careful
studies.
Brain Communications,
Год журнала:
2023,
Номер
6(1)
Опубликована: Дек. 28, 2023
Motor
recovery
is
still
limited
for
people
with
stroke
especially
those
greater
functional
impairments.
In
order
to
improve
outcome,
we
need
understand
more
about
the
mechanisms
underpinning
recovery.
Task-unbiased,
blood
flow-independent
post-stroke
neural
activity
can
be
acquired
from
resting
brain
electrophysiological
recordings
and
offers
substantial
promise
investigate
physiological
mechanisms,
but
behaviourally
relevant
features
of
resting-state
sensorimotor
network
dynamics
have
not
yet
been
identified.
Thirty-seven
subcortical
ischaemic
unilateral
hand
paresis
any
degree
were
longitudinally
evaluated
at
3
weeks
(early
subacute)
12
(late
after
stroke.
Resting-state
magnetoencephalography
clinical
scores
motor
function
recorded
compared
matched
controls.
Magnetoencephalography
data
decomposed
using
a
data-driven
hidden
Markov
model
into
10
time-varying
networks.
People
showed
statistically
significantly
improved
Action
Research
Arm
Test
Fugl-Meyer
upper
extremity
between
(both
Neural
activity
contains
rich
spatiotemporal
structure
that
corresponds
to
cognition.
This
includes
oscillatory
bursting
and
dynamic
span
across
networks
of
brain
regions,
all
which
can
occur
on
timescales
tens
milliseconds.
While
these
processes
be
accessed
through
recordings
imaging,
modeling
them
presents
methodological
challenges
due
their
fast
transient
nature.
Furthermore,
the
exact
timing
duration
interesting
cognitive
events
are
often
a
priori
unknown.
Here,
we
present
OHBA
Software
Library
Dynamics
Toolbox
(osl-dynamics),
Python-based
package
identify
describe
recurrent
dynamics
in
functional
neuroimaging
data
as
At
its
core
machine
learning
generative
models
able
adapt
learn
timing,
well
spatial
spectral
characteristics,
with
few
assumptions.
osl-dynamics
incorporates
state-of-the-art
approaches
be,
have
been,
used
elucidate
wide
range
types,
including
magneto/electroencephalography,
magnetic
resonance
invasive
local
field
potential
recordings,
electrocorticography.
It
also
provides
novel
summary
measures
inform
our
understanding
cognition,
behavior,
disease.
We
hope
will
further
function,
ability
enhance
processes.
Predicting
an
individual's
cognitive
traits
or
clinical
condition
using
brain
signals
is
a
central
goal
in
modern
neuroscience.
This
commonly
done
either
structural
aspects,
such
as
connectivity
cortical
thickness,
aggregated
measures
of
activity
that
average
over
time.
But
these
approaches
are
missing
aspect
function:
the
unique
ways
which
unfolds
One
reason
why
dynamic
patterns
not
usually
considered
they
have
to
be
described
by
complex,
high-dimensional
models;
and
it
unclear
how
best
use
models
for
prediction.
We
here
propose
approach
describes
functional
amplitude
Hidden
Markov
model
(HMM)
combines
with
Fisher
kernel,
can
used
predict
individual
traits.
The
kernel
constructed
from
HMM
mathematically
principled
manner,
thereby
preserving
structure
underlying
model.
show
here,
fMRI
data,
HMM-Fisher
accurate
reliable.
compare
other
prediction
methods,
both
time-varying
time-averaged
connectivity-based
models.
Our
leverages
information
about
has
broad
applications
neuroscience
personalised
medicine.
