Journal of Neural Transmission,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Dec. 30, 2024
Speech
change
is
a
biometric
marker
for
Parkinson's
disease
(PD).
However,
evaluating
speech
variability
across
diverse
languages
challenging.
We
aimed
to
develop
cross-language
algorithm
differentiating
between
PD
patients
and
healthy
controls
using
Taiwanese
Korean
data
set.
recruited
299
347
with
from
Taiwan
Korea.
Participants
underwent
smartphone-based
recordings
during
the
"on"
phase.
Each
participant
performed
various
texts,
while
read
standardized,
fixed-length
article.
short-speech
(≦15
syllables)
long-speech
(>
15
were
combined
dataset.
The
merged
dataset
was
split
into
training
set
(controls
vs.
early-stage
PD)
validation
advanced-stage
evaluate
model's
effectiveness
in
based
on
length.
Numerous
acoustic
linguistic
features
extracted
machine
learning
algorithms
distinguish
controls.
area
under
receiver
operating
characteristic
(AUROC)
curve
calculated
assess
diagnostic
performance.
Random
forest
AdaBoost
classifiers
showed
an
AUROC
0.82
distinguishing
In
cohort,
random
maintained
this
value
(0.90)
discriminating
patients.
model
superior
performance
language
cohort
(AUROC
0.90)
than
either
0.87)
or
0.88)
cohorts
individually.
another
of
<
25
characters,
identify
dropped
0.72
further
limited
ability
discriminate
Leveraging
multifaceted
features,
including
both
characteristics,
could
aid
individuals,
even
different
languages.
Scientific Reports,
Journal Year:
2025,
Volume and Issue:
15(1)
Published: April 4, 2025
Abstract
Parkinson’s
disease,
currently
the
fastest-growing
neurodegenerative
disorder
globally,
has
seen
a
50%
increase
in
cases
within
just
two
years.
As
disease
progression
impairs
speech,
memory,
and
motor
functions
over
time,
early
diagnosis
is
crucial
for
preserving
patients’
quality
of
life.
Although
machine-learning-based
detection
shown
promise
detecting
most
studies
rely
on
single
feature
classification
can
be
error-prone
due
to
variability
symptoms
between
patients.
To
address
this
limitation
we
utilized
mPower
dataset,
which
includes
150,000
samples
across
four
key
biomarkers:
voice,
gait,
tapping,
demographic
data.
From
these
measurements,
extracted
64
features
trained
baseline
Random
Forest
model
select
above
80th
percentile.
For
classification,
designed
simulatable
quantum
support
vector
machine
(qSVM)
that
detects
high-dimensional
patterns,
leveraging
recent
advancements
learning.
With
novel
architecture
run
standard
hardware
rather
than
resource-intensive
computers,
our
achieves
an
accuracy
90%,
F-1
score
0.90,
AUC
0.98—surpassing
benchmark
models.
Utilizing
innovative
framework
built
diverse
set
features,
offers
pathway
accessible
global
screening.
Nursing and Residential Care,
Journal Year:
2025,
Volume and Issue:
unknown, P. 1 - 3
Published: April 28, 2025
Sarah
Jane
Palmer
discusses
an
increase
in
the
prevalence
of
Parkinson's
disease
and
what
interventions
can
be
implemented
to
slow
down
process
ageing,
thereby
potentially
reducing
rate.
