Sensors,
Journal Year:
2022,
Volume and Issue:
23(1), P. 40 - 40
Published: Dec. 21, 2022
The
coronavirus
disease
(COVID-19)
pandemic
was
caused
by
the
SARS-CoV-2
virus
and
began
in
December
2019.
first
reported
Wuhan
region
of
China.
It
is
a
new
strain
that
until
then
had
not
been
isolated
humans.
In
severe
cases,
pneumonia,
acute
respiratory
distress
syndrome,
multiple
organ
failure
or
even
death
may
occur.
Now,
existence
vaccines,
antiviral
drugs
appropriate
treatment
are
allies
confrontation
disease.
present
research
work,
we
utilized
supervised
Machine
Learning
(ML)
models
to
determine
early-stage
symptoms
occurrence.
For
this
purpose,
experimented
with
several
ML
models,
results
showed
ensemble
model,
namely
Stacking,
outperformed
others,
achieving
an
Accuracy,
Precision,
Recall
F-Measure
equal
90.9%
Area
Under
Curve
(AUC)
96.4%.
ACM Transactions on Embedded Computing Systems,
Journal Year:
2024,
Volume and Issue:
23(5), P. 1 - 33
Published: July 26, 2024
Modern
advances
in
machine
learning
(ML)
and
wearable
medical
sensors
(WMSs)
edge
devices
have
enabled
ML-driven
disease
detection
for
smart
healthcare.
Conventional
methods
rely
on
customizing
individual
models
each
its
corresponding
WMS
data.
However,
such
lack
adaptability
to
distribution
shifts
new
task
classification
classes.
In
addition,
they
need
be
rearchitected
retrained
from
scratch
disease.
Moreover,
installing
multiple
ML
an
device
consumes
excessive
memory,
drains
the
battery
faster,
complicates
process.
To
address
these
challenges,
we
propose
DOCTOR,
a
multi-disease
continual
(CL)
framework
based
WMSs.
It
employs
multi-headed
deep
neural
network
(DNN)
replay-style
CL
algorithm.
The
algorithm
enables
continually
learn
missions
which
different
data
distributions,
classes,
tasks
are
introduced
sequentially.
counteracts
catastrophic
forgetting
with
either
preservation
(DP)
method
or
synthetic
generation
(SDG)
module.
DP
preserves
most
informative
subset
of
real
training
previous
exemplar
replay.
SDG
module
probability
generates
generative
replay
while
retaining
privacy.
DNN
DOCTOR
detect
diseases
simultaneously
user
We
demonstrate
DOCTOR’s
efficacy
maintaining
high
accuracy
single
model
various
experiments.
complex
scenarios,
achieves
1.43×
better
average
test
accuracy,
1.25×
F1-score,
0.41
higher
backward
transfer
than
naïve
fine-tuning
framework,
small
size
less
350
KB.
World Journal of Advanced Research and Reviews,
Journal Year:
2024,
Volume and Issue:
22(2), P. 926 - 936
Published: May 16, 2024
Hepatitis
C
is
an
infection
of
the
liver
brought
on
by
HCV
virus.
In
this
condition,
early
diagnosis
challenging
because
delayed
onset
symptoms.
Predicting
well
enough
can
spare
patients
from
permeant
damage.
The
primary
goal
work
to
use
several
machine
learning
methods
forecast
disease
based
widely
available
and
reasonably
priced
blood
test
data
in
order
diagnose
treat
on.
Three
techniques
support
vector
(SVM),
logistic
regression,
decision
tree,
has
been
applied
one
dataset
work.
To
find
a
suitable
approach
for
illness
prediction,
confusion
matrix,
precision,
recall,
F1
score,
accuracy,
receiver
operating
characteristics
(ROC),
performances
different
strategies
have
assessed.
SVM
model's
overall
accuracy
0.92,
highest
among
three
models.
Scientific Reports,
Journal Year:
2025,
Volume and Issue:
15(1)
Published: Feb. 24, 2025
The
problem
of
missing
data
in
sets
is
the
most
important
first
step
to
be
addressed
preprocessing
phase.
Because
incorrect
imputation
increases
error
modeling
phase
and
reduces
prediction
performance
model.
When
it
comes
health,
inevitable
choose
models
that
show
a
higher
success
rate.
In
cases
where
there
data,
machine
learning
may
differ
depending
on
amount
contained
set.
presence
this
high
rate
affects
accuracy
reliability
analysis
studies
because
will
affect
complete
Estimating
filling
very
precisely,
close
its
real
value,
provide
significant
visible
increase
phase,
which
next
stage.
After
imputing
with
an
artificial
intelligence
model
rather
than
random
method,
obvious
trained
filled
classical
methods
such
as
mean
mode.
study,
we
propose
new
algorithm
has
been
tested
many
datasets
address
problems
caused
by
dataset.
aims
impute
values
more
effectively
using
row-based
column-based
techniques
together
cyclically.
takes
into
account
individual
features
overall
structure
features.
proposed
achieved
100%
some
row
3
different
used
study.
Higher
was
compared
other
techniques.
Applied Computational Intelligence and Soft Computing,
Journal Year:
2023,
Volume and Issue:
2023, P. 1 - 17
Published: Nov. 22, 2023
Chronic
kidney
disease
(CKD)
is
a
progressive
condition
characterized
by
the
gradual
deterioration
of
functions,
potentially
leading
to
failure
if
not
promptly
diagnosed
and
treated.
Machine
learning
(ML)
algorithms
have
shown
significant
promise
in
diagnosis,
but
healthcare,
clinical
data
pose
challenges:
missing
values,
noisy
inputs,
redundant
features,
affecting
early-stage
CKD
prediction.
Thus,
this
study
presents
novel,
fully
automated
machine
approach
tackle
these
complexities
incorporating
feature
selection
(FS)
space
reduction
(FSR)
techniques,
substantial
enhancement
model’s
performance.
A
balancing
technique
also
employed
during
preprocessing
address
imbalance
issue
that
commonly
encountered
contexts.
Finally,
for
reliable
classification,
an
ensemble
characteristics-based
classifier
encouraged.
The
effectiveness
our
rigorously
validated
assessed
on
multiple
datasets,
relevancy
strategy
evaluated
real-world
therapeutic
collected
from
Bangladeshi
patients.
establishes
dominance
adaptive
boosting,
logistic
regression,
passive
aggressive
ML
classifiers
with
96.48%
accuracy
forecasting
unseen
data,
particularly
cases.
Furthermore,
FSR
reducing
prediction
time
significantly
revealed.
outstanding
performance
proposed
model
demonstrates
its
addressing
complexity
healthcare
FS
techniques.
This
highlights
potential
as
promising
computer-aided
diagnosis
tool
doctors,
enabling
early
interventions
improving
patient
outcomes.
