International Journal of Infectious Diseases,
Journal Year:
2022,
Volume and Issue:
122, P. 802 - 810
Published: July 22, 2022
ABSTRACT
Objectives
This
study
used
the
long-short-term
memory
(LSTM)
artificial
intelligence
method
to
model
multiple
time
points
of
clinical
laboratory
data,
along
with
demographics
and
comorbidities,
predict
hospital-acquired
acute
kidney
injury
(AKI)
onset
in
patients
COVID-19.
Methods
Montefiore
Health
System
data
consisted
1982
AKI
2857
non-AKI
(NAKI)
hospitalized
COVID-19,
Stony
Brook
Hospital
validation
308
721
NAKI
Demographic,
longitudinal
(3
days
before
onset)
tests
were
analyzed.
LSTM
was
fivefold
cross-validation
(80%/20%
for
training/validation).
Results
The
top
predictors
glomerular
filtration
rate,
lactate
dehydrogenase,
alanine
aminotransferase,
aspartate
C-reactive
protein.
Longitudinal
yielded
marked
improvement
prediction
accuracy
over
individual
points.
inclusion
comorbidities
further
improves
accuracy.
best
an
area
under
curve,
accuracy,
sensitivity,
specificity
be
0.965
±
0.003,
89.57
1.64%,
0.95
0.03,
0.84
0.05,
respectively,
dataset,
0.86
0.01,
83.66
2.53%,
0.66
0.10,
0.89
dataset.
Conclusion
accurately
predicted
approach
could
help
heighten
awareness
complications
identify
early
interventions
prevent
long-term
renal
complications.
Scientific Reports,
Journal Year:
2021,
Volume and Issue:
11(1)
Published: Oct. 26, 2021
Abstract
This
study
investigated
pre-COVID-19
admission
dependency,
discharge
assistive
equipment,
medical
follow-up
recommendation,
and
functional
status
at
hospital
of
non-critically
ill
COVID-19
survivors,
stratified
by
those
with
(N
=
155)
without
162)
in-hospital
rehabilitation.
“Mental
Status”,
intensive-care-unit
(ICU)
Mobility,
modified
Barthel
Index
scores
were
assessed
discharge.
Relative
to
the
non-rehabilitation
patients,
rehabilitation
patients
older,
had
more
comorbidities,
worse
pre-admission
discharged
equipment
supplemental
oxygen,
spent
days
in
hospital,
hospital-acquired
acute
kidney
injury,
respiratory
failure,
referrals
(
p
<
0.05
for
all).
Cardiology,
vascular
medicine,
urology,
endocrinology
amongst
top
referrals.
Functional
many
survivors
abnormal
0.05)
associated
dependency
0.05).
Some
negatively
correlated
age,
hypertension,
coronary
artery
disease,
chronic
psychiatric
anemia,
neurological
disorders
In-hospital
providing
restorative
therapies
assisting
planning
challenging
circumstances.
Knowledge
status,
recommendations
could
enable
appropriate
timely
post-discharge
care.
Follow-up
studies
are
warranted
as
will
likely
have
significant
post-acute
sequela.
Research Square (Research Square),
Journal Year:
2024,
Volume and Issue:
unknown
Published: June 28, 2024
Abstract
Objectives.
This
study
investigated
post
COVID-19
outcomes
of
patients
with
pre-existing
neurological
conditions
up
to
3.5
years
post-infection.
Methods.
retrospective
consisted
1,664
(of
which
1,320
had
been
hospitalized
for
acute
COVID-19)
and
8,985
non-COVID
from
the
Montefiore
Health
System
in
Bronx
(Jan-2016
Jul-2023).
Primary
were
all-cause
mortality
major
adverse
cardiovascular
events
(MACE)
post-COVID-19.
Secondary
depression,
anxiety,
fatigue,
headache,
sleep
disturbances,
altered
mental
status,
dyspnea
Cox
proportional
hazards
model
was
used
calculate
adjusted
hazard
ratios
event
(MACE).
Cumulative
incidence
function
Fine-Gray
sub-distribution
analysis
performed
secondary
outcomes.
Results.
Patients
a
disease
more
likely
die
(adjusted
HR
=
1.92
[CI:1.60,
2.30],
P
<
0.005),
whereas
non-hospitalized
rate
(aHR
1.08
[CI:0.65,
1.81],
0.76),
compared
patients.
(hospitalized
aHR
1.76
[CI:1.53,
2.03],
0.005;
not
COVID-19:
1.50
[CI:1.09,
2.05],
0.01)
experience
MACE
Notably
Blacks
1.49)
Hispanics
1.35)
higher
risk
MACE.
Both
develop
cumulative
disturbance,
(p
0.05).
Conclusions.
who
contracted
have
worse
controls.
Identifying
at-risk
individuals
could
enable
diligent
follow-up.
Journal of Alzheimer s Disease,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Dec. 16, 2024
Although
COVID-19
has
been
linked
to
worse
acute
outcomes
in
patients
with
some
neurodegenerative
disorders,
its
long-term
impact
on
dementia
remains
unclear.
International Journal of Infectious Diseases,
Journal Year:
2022,
Volume and Issue:
122, P. 802 - 810
Published: July 22, 2022
ABSTRACT
Objectives
This
study
used
the
long-short-term
memory
(LSTM)
artificial
intelligence
method
to
model
multiple
time
points
of
clinical
laboratory
data,
along
with
demographics
and
comorbidities,
predict
hospital-acquired
acute
kidney
injury
(AKI)
onset
in
patients
COVID-19.
Methods
Montefiore
Health
System
data
consisted
1982
AKI
2857
non-AKI
(NAKI)
hospitalized
COVID-19,
Stony
Brook
Hospital
validation
308
721
NAKI
Demographic,
longitudinal
(3
days
before
onset)
tests
were
analyzed.
LSTM
was
fivefold
cross-validation
(80%/20%
for
training/validation).
Results
The
top
predictors
glomerular
filtration
rate,
lactate
dehydrogenase,
alanine
aminotransferase,
aspartate
C-reactive
protein.
Longitudinal
yielded
marked
improvement
prediction
accuracy
over
individual
points.
inclusion
comorbidities
further
improves
accuracy.
best
an
area
under
curve,
accuracy,
sensitivity,
specificity
be
0.965
±
0.003,
89.57
1.64%,
0.95
0.03,
0.84
0.05,
respectively,
dataset,
0.86
0.01,
83.66
2.53%,
0.66
0.10,
0.89
dataset.
Conclusion
accurately
predicted
approach
could
help
heighten
awareness
complications
identify
early
interventions
prevent
long-term
renal
complications.
