Revolutionizing Blood Banks: AI-Driven Fingerprint-Blood Group Correlation for Enhanced Safety
Data & Metadata,
Journal Year:
2025,
Volume and Issue:
4, P. 894 - 894
Published: April 7, 2025
Identification
of
a
person
is
central
in
forensic
science,
security,
and
healthcare.
Methods
such
as
iris
scanning
genomic
profiling
are
more
accurate
but
expensive,
time-consuming,
difficult
to
implement.
This
study
focuses
on
the
relationship
between
fingerprint
patterns
ABO
blood
group
biometric
identification
tool.
A
total
200
subjects
were
included
study,
types
(loops,
whorls,
arches)
groups
compared.
Associations
evaluated
with
statistical
tests,
including
chi-square
Pearson
correlation.The
found
that
loops
most
common
pattern
O+
was
prevalent.
Discussion:
Even
though
there
some
associative
pattern,
no
statistically
significant
difference
different
groups.
Overall,
results
indicate
data
do
not
significantly
improve
personal
when
used
conjunction
fingerprinting.Although
shows
weak
correlation,
it
may
emphasize
efforts
multi-modal
based
systems
enhancing
current
systems.
Future
studies
focus
larger
diverse
samples,
possibly
machine
learning
additional
biometrics
methods.
addresses
an
element
ever-changing
nature
fields
science
identification,
highlighting
importance
resilient
analytical
methods
for
identification.
Language: Английский
Classifying Dental Care Providers Through Machine Learning with Features Ranking
Data & Metadata,
Journal Year:
2025,
Volume and Issue:
4, P. 755 - 755
Published: April 7, 2025
This
study
investigates
the
application
of
machine
learning
(ML)
models
for
classifying
dental
providers
into
two
categories—standard
rendering
and
safety
net
clinic
(SNC)
providers—using
a
2018
dataset
24,300
instances
with
20
features.
The
dataset,
characterized
by
high
missing
values
(38.1%),
includes
service
counts
(preventive,
treatment,
exams),
delivery
systems
(FFS,
managed
care),
beneficiary
demographics.
Feature
ranking
methods
such
as
information
gain,
Gini
index,
ANOVA
were
employed
to
identify
critical
predictors,
revealing
treatment-related
metrics
(TXMT_USER_CNT,
TXMT_SVC_CNT)
top-ranked
Twelve
ML
models,
including
k-Nearest
Neighbors
(kNN),
Decision
Trees,
Support
Vector
Machines
(SVM),
Stochastic
Gradient
Descent
(SGD),
Random
Forest,
Neural
Networks,
Boosting,
evaluated
using
10-fold
cross-validation.
Classification
accuracy
was
tested
across
incremental
feature
subsets
derived
from
rankings.
Network
achieved
highest
(94.1%)
all
features,
followed
Boosting
(93.2%)
Forest
(93.0%).
Models
showed
improved
performance
more
features
incorporated,
SGD
ensemble
demonstrating
robustness
data.
highlighted
dominance
treatment
annotation
codes
in
distinguishing
provider
types,
while
demographic
variables
(AGE_GROUP,
CALENDAR_YEAR)
had
minimal
impact.
underscores
importance
selection
enhancing
model
efficiency
accuracy,
particularly
imbalanced
healthcare
datasets.
These
findings
advocate
integrating
feature-ranking
techniques
advanced
algorithms
optimize
classification,
enabling
targeted
resource
allocation
underserved
populations.
Language: Английский
Predicting Blood Type: Assessing Model Performance with ROC Analysis
Data & Metadata,
Journal Year:
2025,
Volume and Issue:
4, P. 895 - 895
Published: April 9, 2025
Introduction:
Personal
identification
is
a
critical
aspect
of
forensic
sciences,
security,
and
healthcare.
While
conventional
biometrics
systems
such
as
DNA
profiling
iris
scanning
offer
high
accuracy,
they
are
time-consuming
costly.
Objectives:
This
study
investigates
the
relationship
between
fingerprint
patterns
ABO
blood
group
classification
to
explore
potential
correlations
these
two
traits.Methods:
The
analyzed
200
individuals,
categorizing
their
fingerprints
into
three
types:
loops,
whorls,
arches.
Blood
was
also
recorded.
Statistical
analysis,
including
chi-square
Pearson
correlation
tests,
used
assess
associations
groups.Results:
Loops
were
most
common
pattern,
while
O+
prevalent
among
participants.
analysis
revealed
no
significant
groups
(p
>
0.05),
suggesting
that
traits
independent.Conclusions:
Although
showed
limited
groups,
it
highlights
importance
future
research
using
larger
more
diverse
populations,
incorporating
machine
learning
approaches,
integrating
multiple
biometric
signals.
contributes
science
by
emphasizing
need
for
rigorous
protocols
comprehensive
investigations
in
personal
identification.
Language: Английский
Machine Learning-Based Quantification of Vesicoureteral Reflux with Enhancing Accuracy and Efficiency
Data & Metadata,
Journal Year:
2025,
Volume and Issue:
4, P. 756 - 756
Published: March 19, 2025
Vesicoureteral
reflux
(VUR)
is
traditionally
assessed
using
subjective
grading
systems,
leading
to
variability
in
diagnosis.
This
study
explores
the
potential
of
machine
learning
enhance
diagnostic
accuracy
by
analysing
voiding
cystourethrogram
(VCUG)
images.
The
objective
develop
predictive
models
that
provide
an
and
consistent
approach
VUR
classification.
A
total
113
VCUG
images
were
reviewed,
with
experts
them
based
on
severity.
Nine
distinct
image
features
selected
build
six
models,
which
evaluated
'leave-one-out'
cross-validation.
analysis
identified
renal
calyces’
deformation
patterns
as
key
indicators
high-grade
VUR.
models—Logistic
Regression,
Tree,
Gradient
Boosting,
Neural
Network,
Stochastic
Descent—achieved
precise
classifications
no
false
positives
or
negatives.
High
sensitivity
subtle
characteristic
different
grades
was
confirmed
substantial
Area
Under
Curve
(AUC)
values.
demonstrates
can
address
limitations
assessments,
offering
a
more
reliable
standardized
system.
findings
highlight
significance
predictor
severe
cases.
Future
research
should
focus
refining
methodologies,
exploring
additional
features,
expanding
dataset
model
clinical
applicability.
Language: Английский