Frontiers in Immunology,
Год журнала:
2024,
Номер
15
Опубликована: Сен. 3, 2024
Background
Radiotherapy
(RT)
is
a
critical
component
of
treatment
for
locally
advanced
rectal
cancer
(LARC),
though
patient
response
varies
significantly.
The
variability
in
outcomes
partly
due
to
the
resistance
conferred
by
stem
cells
(CSCs)
and
tumor
immune
microenvironment
(TiME).
This
study
investigates
role
EIF5A
radiotherapy
its
impact
on
CSCs
TiME.
Methods
Predictive
models
preoperative
(preRT)
were
developed
using
machine
learning,
identifying
as
key
gene
associated
with
radioresistance.
expression
was
analyzed
via
bulk
RNA-seq
single-cell
(scRNA-seq).
Functional
assays
vivo
experiments
validated
EIF5A’s
radioresistance
TiME
modulation.
Results
significantly
upregulated
radioresistant
colorectal
(CRC)
tissues.
knockdown
CRC
cell
lines
reduced
viability,
migration,
invasion
after
radiation,
increased
radiation-induced
apoptosis.
Mechanistically,
promoted
(CSC)
characteristics
through
Hedgehog
signaling
pathway.
Analysis
revealed
that
radiation-resistant
group
had
an
immune-desert
phenotype,
characterized
low
infiltration.
In
showed
led
infiltration
CD8+
T
M1
macrophages,
decreased
M2
macrophages
Tregs
following
radiation
therapy,
thereby
enhancing
response.
Conclusion
contributes
promoting
CSC
traits
pathway
modulating
immune-suppressive
state.
Targeting
could
enhance
sensitivity
improve
responses,
offering
potential
therapeutic
strategy
optimize
patients.
Journal of Electrical Systems and Information Technology,
Год журнала:
2024,
Номер
11(1)
Опубликована: Июль 16, 2024
Abstract
The
past
few
years
have
seen
an
emergence
of
interest
in
examining
the
significance
machine
learning
(ML)
medical
field.
Diseases,
health
emergencies,
and
disorders
may
now
be
identified
with
greater
accuracy
because
technological
advancements
advances
ML.
It
is
essential
especially
to
diagnose
individuals
chronic
diseases
(CD)
as
early
possible.
Our
study
has
focused
on
analyzing
ML’s
applicability
predict
CD,
including
cardiovascular
disease,
diabetes,
cancer,
liver,
neurological
disorders.
This
offered
a
high-level
summary
previous
research
ML-based
approaches
for
predicting
CD
some
instances
their
applications.
To
wrap
things
up,
we
compared
results
obtained
by
various
studies
methodologies
well
tools
employed
researchers.
factors
or
parameters
that
are
responsible
improving
model
different
works
also
identified.
For
identifying
significant
features,
most
authors
variety
strategies,
where
least
absolute
shrinkage
selection
(LASSO),
minimal-redundancy-maximum-relevance
(mRMR),
RELIEF
extensively
used
methods.
wide
range
ML
approaches,
support
vector
(SVM),
random
forest
(RF),
decision
tree
(DT),
naïve
Bayes
(NB),
etc.,
been
widely
used.
Also,
several
deep
techniques
hybrid
models
create
prediction
models,
resulting
efficient
reliable
clinical
decision-making
models.
benefit
whole
healthcare
system,
our
suggestions
enhancing
CD.
Diabetes Metabolic Syndrome and Obesity,
Год журнала:
2025,
Номер
Volume 18, С. 267 - 282
Опубликована: Янв. 1, 2025
Type
2
diabetes
mellitus
(T2DM)
is
associated
with
an
increased
risk
of
non-Hodgkin
lymphoma
(NHL),
but
the
underlying
mechanisms
remain
unclear.
This
study
aimed
to
identify
potential
biomarkers
and
elucidate
molecular
co-pathogenesis
T2DM
NHL.
Microarray
datasets
NHL
were
downloaded
from
Gene
Expression
Omnibus
database.
Subsequently,
a
protein-protein
interaction
network
was
constructed
based
on
common
differentially
expressed
genes
(DEGs)
between
explore
regulatory
interactions.
Functional
analyses
performed
mechanisms.
Topological
analysis
machine
learning
algorithms
applied
refine
hub
gene
selection.
Finally,
quantitative
real-time
polymerase
chain
reaction
validate
in
clinical
samples.
Intersection
DEGs
identified
81
shared
genes.
suggested
that
immune-related
pathways
played
significant
role
three
genes:
GZMM,
HSPG2,
SERPING1.
Correlation
revealed
correlations
these
immune
cells,
underscoring
importance
dysregulation
pathogenesis.
The
expression
successfully
validated
pivotal
as
key
contributors.
These
findings
provide
insight
into
complex
interplay
IEEE Journal of Translational Engineering in Health and Medicine,
Год журнала:
2024,
Номер
12, С. 569 - 579
Опубликована: Янв. 1, 2024
Brain
microstructural
changes
already
occur
in
the
earliest
phases
of
Alzheimer's
disease
(AD)
as
evidenced
diffusion
magnetic
resonance
imaging
(dMRI)
literature.
This
study
investigates
potential
novel
dMRI
Apparent
Measures
Using
Reduced
Acquisitions
(AMURA)
markers
for
capturing
such
tissue
modifications.Tract-based
spatial
statistics
(TBSS)
and
support
vector
machines
(SVMs)
based
on
different
measures
were
exploited
to
distinguish
between
amyloid-beta/tau
negative
(A
$\beta
$
-/tau-)
A
+/tau+
or
+/tau-
subjects.
Moreover,
eXplainable
Artificial
Intelligence
(XAI)
was
used
highlight
most
influential
features
SVMs
classifications
validate
results
by
seeing
explanations'
recurrence
across
methods.TBSS
analysis
revealed
significant
differences
-/tau-
other
groups
line
with
The
best
SVM
classification
performance
reached
an
accuracy
0.73
using
advanced
compared
more
standard
ones.
explainability
suggested
results'
stability
central
role
cingulum
show
early
sign
AD.By
relying
XAI
interpretation
outcomes,
AMURA
indices
can
be
considered
viable
amyloid
tau
pathology.
Clinical
impact:
pre-clinical
research
timely
AD
diagnosis
acquiring
clinically
feasible
dMR
images,
advantages
invasive
methods
employed
nowadays.
ABSTRACT
Background
Advances
in
imaging
technology
have
enhanced
the
detection
of
pulmonary
nodules.
However,
determining
malignancy
often
requires
invasive
procedures
or
repeated
radiation
exposure,
underscoring
need
for
safer,
noninvasive
diagnostic
alternatives.
Analyzing
exhaled
volatile
organic
compounds
(VOCs)
shows
promise,
yet
its
effectiveness
assessing
nodules
remains
underexplored.
Methods
Employing
a
prospective
study
design
from
June
2023
to
January
2024
at
Affiliated
Hospital
Yangzhou
University,
we
assessed
using
Mayo
Clinic
model
and
collected
breath
samples
alongside
lifestyle
health
examination
data.
We
applied
five
machine
learning
(ML)
algorithms
develop
predictive
models
which
were
evaluated
area
under
curve
(AUC),
sensitivity,
specificity,
other
relevant
metrics.
Results
A
total
267
participants
enrolled,
including
210
with
low‐risk
57
moderate‐risk
Univariate
analysis
identified
11
VOCs
associated
nodule
malignancy,
two
factors
(smoke
index
sites
tobacco
smoke
inhalation)
one
clinical
metric
(nodule
diameter)
as
independent
predictors
The
logistic
regression
integrating
data
achieved
an
AUC
0.91
(95%
CI:
0.8611–0.9658),
while
random
forest
incorporating
0.99
0.974–1.00).
Calibration
curves
indicated
strong
concordance
between
predicted
observed
risks.
Decision
confirmed
net
benefit
these
over
traditional
methods.
nomogram
was
developed
aid
clinicians
based
on
VOCs,
lifestyle,
Conclusions
integration
ML
biomarkers
provides
robust
framework
assessment
These
offer
safer
alternative
methods
may
enhance
early
management
Further
validation
through
larger,
multicenter
studies
is
necessary
establish
their
generalizability.
Trial
Registration:
Number
ChiCTR2400081283
Clinical and Experimental Obstetrics & Gynecology,
Год журнала:
2025,
Номер
52(1)
Опубликована: Янв. 13, 2025
Background:
The
prognosis
of
patients
with
early
diagnosis
malignant
endometrial
lesions
is
good.
We
aimed
to
identify
benign
and
in
tissue,
explore
effective
methods
for
assisting
diagnosis,
improve
the
accuracy
precision
identifying
lesions.
Methods:
1142
ultrasound
radiomics
18
clinical
features
from
1254
were
analyzed,
which
36
selected
machine
learning.
sketched
region
interest
(ROI)
abnormalities
on
images.
Then,
extracted.
Six
common
learning
algorithms,
including
Support
Vector
Machine
(SVM),
Logistic
Regression,
Decision
Tree,
Random
Forest,
Gradient
Boosting
k-Nearest
Neighbors,
employed
changes
tissue.
Cross-validation
grid
search
techniques
hyperparameter
tuning
utilized
obtain
best
model
performance.
Accuracy,
precision,
sensitivity,
F1-scores,
area
under
curve
(AUC),
cross-validation
average
score
bootstrap
also
used
evaluate
algorithm
performance,
classification
accuracy,
generalization
capability.
Results:
combined
21
characteristics
15
develop
validate
six
algorithms.
After
internal
validation,
models
Forest
models,
89%,
93%,
sensitivity
97%,
F1-score
95%,
AUC
as
well
a
10-fold
95%
94%,
implying
flawless
test
set.
Conclusions:
identified
or
And
algorithms
have
demonstrated
excellent
performance
This
significant
enhancing
diagnostic
improving
treatment
outcomes
long-term
management.