Advances in human and social aspects of technology book series,
Год журнала:
2024,
Номер
unknown, С. 137 - 166
Опубликована: Окт. 18, 2024
As
the
global
population
ages,
ensuring
emotional
well-being
of
older
adults
is
a
critical
aspect
healthcare
systems.
Emotional
deeply
intertwined
with
physical
and
mental
health,
particularly
in
facing
complex
health
challenges
such
as
falls,
nutritional
deficiencies,
chronic
conditions
like
cancer
kidney
disease
(CKD),
concerns
loneliness.
In
this
chapter,
we
explore
how
artificial
intelligence
(AI)
can
enhance
by
promoting
(EI)
improving
outcomes
for
adults.
The
discussion
integrates
key
aspects
including
falls
prevention,
nutrition,
hydration,
vitamin
B12
deficiency,
ageing
place,
loneliness,
care,
pharmacokinetics,
moving
handling.
We
examine
transformative
potential
AI
technologies
addressing
these
issues,
offering
real-time
personalized
care
interventions.
By
incorporating
into
elder
create
holistic
systems
that
support
emotional,
mental,
health.
Journal of Medical Internet Research,
Год журнала:
2025,
Номер
27, С. e68030 - e68030
Опубликована: Апрель 30, 2025
Perinatal
depression
and
anxiety
significantly
impact
maternal
infant
health,
potentially
leading
to
severe
outcomes
like
preterm
birth
suicide.
Aboriginal
women,
despite
their
resilience,
face
elevated
risks
due
the
long-term
effects
of
colonization
cultural
disruption.
The
Baby
Coming
You
Ready
(BCYR)
model
care,
centered
on
a
digitized,
holistic,
strengths-based
assessment,
was
co-designed
address
these
challenges.
successful
BCYR
pilot
demonstrated
its
ability
replace
traditional
risk-based
screens.
However,
some
health
professionals
still
overrely
psychological
risk
scores,
often
overlooking
contextual
circumstances
mothers,
strengths,
mitigating
protective
factors.
This
highlights
need
for
new
tools
improve
clinical
decision-making.
We
explored
different
explainable
artificial
intelligence
(XAI)-powered
machine
learning
techniques
developing
culturally
informed,
predictive
modeling
perinatal
distress
among
mothers.
identifies
evaluates
influential
factors
while
offering
transparent
explanations
AI-driven
decisions.
used
deidentified
data
from
293
mothers
who
participated
in
program
between
September
2021
June
2023
at
6
care
services
Perth
regional
Western
Australia.
original
dataset
includes
variables
spanning
factors,
life
events,
worries,
relationships,
childhood
experiences,
family
domestic
violence,
substance
use.
After
applying
feature
selection
expert
input,
20
were
chosen
as
predictors.
Kessler-5
scale
an
indicator
distress.
Several
models,
including
random
forest
(RF),
CatBoost
(CB),
light
gradient-boosting
(LightGBM),
extreme
gradient
boosting
(XGBoost),
k-nearest
neighbor
(KNN),
support
vector
(SVM),
(EBM),
developed
compared
performance.
To
make
black-box
interpretable,
post
hoc
explanation
Shapley
additive
local
interpretable
model-agnostic
applied.
EBM
outperformed
other
models
(accuracy=0.849,
95%
CI
0.8170-0.8814;
F1-score=0.771,
0.7169-0.8245;
area
under
curve=0.821,
0.7829-0.8593)
followed
by
RF
(accuracy=0.829,
0.7960-0.8617;
F1-score=0.736,
0.6859-0.7851;
curve=0.795,
0.7581-0.8318).
Explanations
EBM,
explanations,
identified
consistent
patterns
key
questions
related
"Feeling
Lonely,"
"Blaming
Herself,"
"Makes
Family
Proud,"
"Life
Not
Worth
Living,"
"Managing
Day-to-Day."
At
individual
level,
where
responses
are
highly
personal,
XAI
provided
case-specific
insights
through
visual
representations,
distinguishing
illustrating
predictions.
study
shows
potential
XAI-driven
predict
provide
clear,
human-interpretable
how
important
interact
influence
outcomes.
These
may
help
more
non-biased
decisions
mental
screenings.
Diagnostics,
Год журнала:
2024,
Номер
14(21), С. 2385 - 2385
Опубликована: Окт. 25, 2024
Depression
is
a
pervasive
mental
health
condition,
particularly
affecting
older
adults,
where
early
detection
and
intervention
are
essential
to
mitigate
its
impact.
This
study
presents
an
explainable
multi-layer
dynamic
ensemble
framework
designed
detect
depression
assess
severity,
aiming
improve
diagnostic
precision
provide
insights
into
contributing
factors.
Suicide
is
one
of
the
major
causes
death
globally.
Analysis
social
media
posts
and
in-depth
insights
show
that
some
people
have
suicide
ideas.
In
order
to
save
more
lives,
it
crucial
comprehend
behavior
suicidal
attempters.
However,
identifying
explaining
thoughts
poses
a
significant
challenge
in
psychiatry.
Additionally,
analysing
complex
procedure
involving
several
variables
based
on
individual's
preferences
data
type.
Although
traditional
methods
been
utilized
identify
clinical
factors
for
ideation
detection
(SID),
these
models
often
lack
interpretability
understanding.
Therefore,
primary
aim
this
research
apply
deep
learning
(DL)
machine
(ML)
techniques
such
as
BERT,
LSTM,
BiLSTM,
RF,
SVM,
GaussianNB,
LR,
KNeighbors
blending
with
interpretable
LIME
SHAP
provide
valuable
into
importance
different
features
make
transparent
SID
process.
The
experiments
were
conducted
publicly
available
dataset
comprising
24,101
posts,
categorized
either
or
non-suicidal.
implemented
method
brings
about
enhancements
performance
comparison.
A
comparison
all
measures
reveals
LSTM
model
particularly
good
at
processing
classifying
textual
data,
higher
accuracy,
precision,
recall,
AUC
scores
than
other
tested.
Sensors,
Год журнала:
2023,
Номер
24(1), С. 164 - 164
Опубликована: Дек. 27, 2023
Digital
health
applications
using
Artificial
Intelligence
(AI)
are
a
promising
opportunity
to
address
the
widening
gap
between
available
resources
and
mental
needs
globally.
Increasingly,
passively
acquired
data
from
wearables
augmented
with
carefully
selected
active
depressed
individuals
develop
Machine
Learning
(ML)
models
of
depression
based
on
mood
scores.
However,
most
ML
black
box
in
nature,
hence
outputs
not
explainable.
Depression
is
also
multimodal,
reasons
for
may
vary
significantly
individuals.
Explainable
personalised
will
thus
be
beneficial
clinicians
determine
main
features
that
lead
decline
state
individual,
enabling
suitable
therapy.
This
currently
lacking.
Therefore,
this
study
presents
methodology
developing
accurate
Deep
(DL)-based
predictive
depression,
along
novel
methods
identifying
key
facets
exacerbation
depressive
symptoms.
We
illustrate
our
approach
by
an
existing
multimodal
dataset
containing
longitudinal
Ecological
Momentary
Assessments
lifestyle
neurocognitive
assessments
14
mild
moderately
participants
over
one
month.
classification-
regression-based
DL
predict
participants’
current
scores—a
discrete
score
given
participant
severity
their
The
trained
inside
eight
different
evolutionary-algorithm-based
optimisation
schemes
optimise
model
parameters
maximum
performance.
A
five-fold
cross-validation
scheme
used
verify
model’s
performance
against
10
classical
ML-based
models,
error
as
low
6%
some
participants.
use
best
process
extract
indicators,
SHAP,
ALE
Anchors
explainable
AI
literature
explain
why
certain
predictions
made
how
they
affect
mood.
These
feature
insights
can
assist
professionals
incorporating
interventions
into
individual’s
treatment
regimen.
Research Square (Research Square),
Год журнала:
2024,
Номер
unknown
Опубликована: Авг. 2, 2024
Abstract
Background
Perinatal
mental
health
significantly
affects
mothers,
infants,
and
families.
Despite
their
resilience
strengths,
Aboriginal
mothers
experience
disproportionate
physical
disparities.
These
result
from
historical
ongoing
impacts
of
colonization
the
resultant
complex
trauma.
Conventional
approaches
to
perinatal
care
present
many
barriers
for
who
frequently
feel
disengaged,
apprehensive
unsafe.
Current
score-based
risk-screening
practices
that
algorithmically
drive
referrals,
further
ingrain
fears
including
culturally
biased
judgments
child
removal.
The
Baby
Coming
You
Ready
(BCYR)
model
centred
around
a
digitised,
holistic,
strengths-based
assessment,
was
co-designed
address
these
barriers.
recent
successful
pilot
demonstrated
BCYR
effectively
replaced
all
current
risk-based
screens.
However,
professionals
disproportionately
rely
on
psychological
risk
scores,
overlooking
contextual
circumstances
cultural
strengths
mitigating
protective
factors.
Methods
To
this
singular
reliance
screening
psychometrics
whilst
supporting
strengthened
considered
clinical
we
propose
sensitive
eXplainable
AI
(XAI)
solution.
It
combines
XAI
with
lived
experience,
knowledge
wisdom
generate
prediction
support
being
screened.
solution
can
identify,
prioritise,
weigh
both
maternal
factors,
quantify
relative
mental-health
well-being
at
group
individual
levels.
Results
Different
machine
learning
algorithms,
Random
Forest,
K-nearest
neighbour,
vector
machine,
alongside
glassbox
Explainable
Boosting
Machine
(EBM)
models,
were
trained
real
life
de-identified
data
generated
during
pilot.
Additionally,
techniques
like
SHAP
LIME
are
utilised
interpretability
black
box
models.
show
EBM
demonstrates
superior
performance
in
prediction,
an
accuracy
0.849,
F1
score
0.771
AUC
0.821.
Global
explanations
across
entire
dataset
local
cases,
achieved
through
different
methods,
compared
showed
similar
stable
results.
Conclusions
This
study
potential
enhance
professionals'
capability
responsive
reasoning
improve
strengthen
outcomes
women.
International Journal of Advanced Computer Science and Applications,
Год журнала:
2024,
Номер
15(7)
Опубликована: Янв. 1, 2024
This
study
proposed
a
novel
approach
to
handle
mental
health,
particularly,
depression
among
college
students,
called
CRADDS
A
Comprehensive
Real-time
Adaptive
Depression
Detection
System.
The
combined
advanced
tensor
fusion
networks
which
is
able
analyze
emotions
using
audio,
text
and
video
data
more
accurately,
this
possible
due
the
strength
of
deep
learning
multimodal
approaches.
system
constructed
with
hybrid
algorithm
framework
that
combines
SVM
(Support
Vector
Machines),
CNN
(Convolutional
Neural
Network)
(Bidirectional
Long-Term
Short-Term
Memory)
BiLSTM
techniques.
To
address
limitations
identified
in
earlier
research,
increasing
its
feature
set
effective
machine
algorithms
reduce
false
positives
negatives.
Further,
it
includes
IoT
devices
collect
real
time
from
various
range
public
private
sources.
symptoms
may
be
continuously
monitored
time,
helps
identify
depressions
early
stages
guaranteed
perfect
well-being
students.
Additionally,
model
has
ability
adjust
based
on
interaction
features,
provide
psychological
support
automatic
responses
observed
verbal
nonverbal
clues.
Experiments
show
obtained
an
impressive
accuracy
features
text,
audio
video,
when
compared
existing
models.
Overall,
useful
tool
for
health
professionals
educational
institutions
because
not
only
identifies
but
also
treat
earlier,
guarantees
good
academic
scores
general
well-being.
validation
increases
63.04%
86.08%
higher
than
model.
Advances in computational intelligence and robotics book series,
Год журнала:
2024,
Номер
unknown, С. 219 - 262
Опубликована: Дек. 23, 2024
This
chapter
explores
the
incorporation
of
artificial
intelligence
(AI)
into
mental
health
care,
with
a
particular
focus
on
managing
depression.
AI
has
significantly
enhanced
promotion,
detection,
diagnosis,
treatment,
and
monitoring
depression
by
leveraging
technologies
such
as
machine
learning,
natural
language
processing,
wearable
devices.
also
discusses
various
AI-driven
approaches,
including
analysis
questionnaires,
medical
records,
social
media,
speech
data,
electroencephalogram,
magnetic
resonance
imaging,
chatbots,
virtual
reality,
face
analysis,
robots,
multimodal
methods,
Each
these
offers
unique
benefits,
increased
accuracy
in
detecting
depression,
personalized
treatment
plans,
continuous
patient
monitoring.
However,
challenges
linked
to
health,
data
privacy
issues,
biases
algorithms,
complexity
human
emotions.
The
concludes
highlighting
opportunities
future
research
directions
innovation
for
enhancing
care.