Identifiability and model selection frameworks for models of high-grade glioma response to chemoradiation
Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences,
Год журнала:
2025,
Номер
383(2293)
Опубликована: Апрель 2, 2025
We
have
developed
a
family
of
biology-based
mathematical
models
high-grade
glioma
(HGG),
capturing
the
key
features
tumour
growth
and
response
to
chemoradiation.
now
seek
quantify
accuracy
parameter
estimation
determine,
when
given
virtual
patient
cohort,
which
model
was
used
generate
tumours.
In
this
way,
we
systematically
test
both
identifiability.
Virtual
patients
are
generated
from
unique
parameters
whose
dynamics
determined
by
family.
then
assessed
ability
recover
select
tumour.
evaluated
predictions
using
selected
at
four
weeks
post-chemoradiation.
observed
median
errors
0.04%
72.96%.
Our
selection
framework
that
data
in
82%
cases.
Finally,
predicted
tumours
resulting
low
error
voxel-level
(concordance
correlation
coefficient
(CCC)
ranged
0.66
0.99)
global
level
(percentage
total
cellularity
−12.35%
0.07%).
These
results
demonstrate
reliability
our
identify
most
appropriate
under
noisy
conditions
expected
clinical
setting.
This
article
is
part
theme
issue
'Uncertainty
quantification
for
healthcare
biological
systems
(Part
2)'.
Язык: Английский
Assessing the role of model choice in parameter identifiability of cancer treatment efficacy
Frontiers in Applied Mathematics and Statistics,
Год журнала:
2025,
Номер
11
Опубликована: Март 24, 2025
Several
mathematical
models
are
commonly
used
to
describe
cancer
growth
dynamics.
Fitting
of
these
experimental
data
has
not
yet
determined
which
particular
model
best
describes
growth.
Unfortunately,
choice
is
known
drastically
alter
the
predictions
both
future
tumor
and
effectiveness
applied
treatment.
Since
there
growing
interest
in
using
help
predict
chemotherapy,
we
need
determine
if
affects
estimates
chemotherapy
efficacy.
Here,
simulate
an
vitro
study
by
creating
synthetic
treatment
each
seven
fit
sets
other
(“wrong”)
models.
We
estimate
ε
max
(the
maximum
efficacy
drug)
IC
50
drug
concentration
at
half
effect
achieved)
effort
whether
use
incorrect
changes
parameters.
find
that
largely
weakly
practically
identifiable
no
matter
generate
or
data.
The
more
likely
be
identifiable,
but
sensitive
model,
showing
poor
identifiability
when
Bertalanffy
either
Язык: Английский
CONVERGENCE, SAMPLING AND TOTAL ORDER ESTIMATOR EFFECTS ON PARAMETER ORTHOGONALITY IN GLOBAL SENSITIVITY ANALYSIS
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2024,
Номер
unknown
Опубликована: Фев. 29, 2024
A
bstract
Dynamical
system
models
typically
involve
numerous
input
parameters
whose
“effects”
and
orthogonality
need
to
be
quantified
through
sensitivity
analysis,
identify
inputs
contributing
the
greatest
uncertainty.
Whilst
prior
art
has
compared
total-order
estimators’
role
in
recovering
“true”
effects,
assessing
their
ability
recover
robust
parameter
for
use
identifiability
metrics
not
been
investigated.
In
this
paper,
we
perform:
(i)
an
assessment
using
a
different
class
of
numerical
representing
cardiovascular
system,
(ii)
wider
evaluation
sampling
methodologies
interactions
with
estimators,
(iii)
investigation
consequences
permuting
estimators
on
orthogonality,
(iv)
study
sample
convergence
resampling,
(v)
whether
positive
outcomes
are
sustained
when
model
dimensionality
increases.
Our
results
indicate
that
Jansen
or
Janon
display
efficient
minimum
uncertainty
coupled
Sobol
lattice
rule
methods,
making
them
prime
choices
calculating
influence.
This
reveals
global
analysis
is
driven.
Unconverged
indices
subject
error
therefore
true
influence
recovered.
importantly
clarifies
estimator
methodology
by
reducing
associated
ambiguities,
defining
novel
practices
modelling
life
sciences.
Research
Highlights
We
conduct
heuristic
utilising
2
physiologically
intuitive,
highly
nonlinear
stiff,
lumped
models.
The
emerge
as
optimal
they
insensitive
measurement
types.
prove
have
most
rates
total
order
indices.
rate
appears
decisive
its
truthfully
uniformly
orthogonality.
methods
provide
putative
best
practice
practical
investigations.
Author
Summary
gain
new
insight
into
biological
systems
one
often
uses
mathematical
predict
possible
responses
from
interest.
One
vital
step
such
knowledge
response
given
change
provided
model.
Utilising
two
non-linear
stiff
test
cases
investigate
effects
made
quantifying
Leveraging
solving
able
show
truly
quantify
set
outputs
must
ensure
converged
estimates
Without
this,
identifying
become
uncertain,
clinically,
non
patient
specific.
detailed
provides
workflow
advice
thus
ensuring
interpretation
inputs.
Язык: Английский
A comparative analysis of 2D and 3D experimental data for the identification of the parameters of computational models
Scientific Reports,
Год журнала:
2023,
Номер
13(1)
Опубликована: Сен. 22, 2023
Computational
models
are
becoming
an
increasingly
valuable
tool
in
biomedical
research.
Their
accuracy
and
effectiveness,
however,
rely
on
the
identification
of
suitable
parameters
appropriate
validation
in-silico
framework.
Both
these
steps
highly
dependent
experimental
model
used
as
a
reference
to
acquire
data.
Selecting
most
framework
thus
becomes
key,
together
with
analysis
effect
combining
results
from
different
models,
common
practice
often
necessary
due
limited
data
availability.
In
this
work,
same
ovarian
cancer
cell
growth
metastasis,
was
calibrated
datasets
acquired
traditional
2D
monolayers,
3D
culture
or
combination
two.
The
comparison
between
sets
obtained
conditions,
corresponding
simulated
behaviours,
is
presented.
It
provides
for
study
development
computational
systems.
This
work
also
set
general
guidelines
comparative
testing
selection
protocols
be
parameter
optimization
models.
Язык: Английский
Personalized Plasma Medicine for Cancer: Transforming Treatment Strategies with Mathematical Modeling and Machine Learning Approaches
Applied Sciences,
Год журнала:
2023,
Номер
14(1), С. 355 - 355
Опубликована: Дек. 30, 2023
Plasma
technology
shows
tremendous
potential
for
revolutionizing
oncology
research
and
treatment.
Reactive
oxygen
nitrogen
species
electromagnetic
emissions
generated
through
gas
plasma
jets
have
attracted
significant
attention
due
to
their
selective
cytotoxicity
towards
cancer
cells.
To
leverage
the
full
of
medicine,
researchers
explored
use
mathematical
models
various
subsets
or
approaches
within
machine
learning,
such
as
reinforcement
learning
deep
learning.
This
review
emphasizes
application
advanced
algorithms
in
adaptive
system,
paving
way
precision
dynamic
Realizing
techniques
medicine
requires
efforts,
data
sharing,
interdisciplinary
collaborations.
Unraveling
complex
mechanisms,
developing
real-time
diagnostics,
optimizing
will
be
crucial
harnessing
true
power
oncology.
The
integration
personalized
therapies,
alongside
AI
diagnostic
sensors,
presents
a
transformative
approach
treatment
with
improve
outcomes
globally.
Язык: Английский
A joint physics and radiobiology DREAM team vision – Towards better response prediction models to advance radiotherapy
Radiotherapy and Oncology,
Год журнала:
2024,
Номер
196, С. 110277 - 110277
Опубликована: Апрель 25, 2024
Radiotherapy
developed
empirically
through
experience
balancing
tumour
control
and
normal
tissue
toxicities.
Early
simple
mathematical
models
formalized
this
practical
knowledge
enabled
effective
cancer
treatment
to
date.
Remarkable
advances
in
technology,
computing,
experimental
biology
now
create
opportunities
incorporate
into
enhanced
computational
models.
The
ESTRO
DREAM
(Dose
Response,
Experiment,
Analysis,
Modelling)
workshop
brought
together
experts
across
disciplines
pursue
the
vision
of
personalized
radiotherapy
for
optimal
outcomes
advanced
modelling.
ultimate
is
leveraging
quantitative
dynamically
during
therapy
ultimately
achieve
truly
adaptive
biologically
guided
at
population
as
well
individual
patient-based
levels.
This
requires
generation
that
inform
response-based
adaptations,
individually
optimized
delivery
enable
biological
monitoring
provide
decision
support
clinicians.
goal
expanding
can
drive
realization
outcomes.
position
paper
provides
their
propositions
describe
how
innovations
biology,
physics,
mathematics,
data
science
including
AI
could
improve
predictions.
It
consolidates
team's
consensus
on
scientific
priorities
organizational
requirements.
Scientifically,
it
stresses
need
rigorous,
multifaceted
model
development,
comprehensive
validation
clinical
applicability
significance.
Organizationally,
reinforces
prerequisites
interdisciplinary
research
collaboration
between
physicians,
medical
physicists,
radiobiologists,
scientists
throughout
development.
Solely
by
a
shared
understanding
needs,
mechanisms,
methods,
more
informed
be
created.
Future
environment
must
facilitate
integrative
method
operation
multiple
disciplines.
Язык: Английский
Practical parameter identifiability and handling of censored data with Bayesian inference in mathematical tumour models
npj Systems Biology and Applications,
Год журнала:
2024,
Номер
10(1)
Опубликована: Авг. 14, 2024
Abstract
Mechanistic
mathematical
models
(MMs)
are
a
powerful
tool
to
help
us
understand
and
predict
the
dynamics
of
tumour
growth
under
various
conditions.
In
this
work,
we
use
5
MMs
with
an
increasing
number
parameters
explore
how
certain
(often
overlooked)
decisions
in
estimating
from
data
experimental
affect
outcome
analysis.
particular,
propose
framework
for
including
volume
measurements
that
fall
outside
upper
lower
limits
detection,
which
normally
discarded.
We
demonstrate
excluding
censored
results
overestimation
initial
MM-predicted
volumes
prior
first
measurements,
underestimation
carrying
capacity
beyond
latest
measurable
time
points.
show
way
choice
MM
can
impact
posterior
distributions,
illustrate
reporting
most
likely
their
95%
credible
interval
lead
confusing
or
misleading
interpretations.
hope
work
will
encourage
others
carefully
consider
choices
made
parameter
estimation
adopt
approaches
put
forward
herein.
Язык: Английский
Structural and practical identifiability of contrast transport models for DCE-MRI
PLoS Computational Biology,
Год журнала:
2024,
Номер
20(5), С. e1012106 - e1012106
Опубликована: Май 15, 2024
Contrast
transport
models
are
widely
used
to
quantify
blood
flow
and
in
dynamic
contrast-enhanced
magnetic
resonance
imaging.
These
analyze
the
time
course
of
contrast
agent
concentration,
providing
diagnostic
prognostic
value
for
many
biological
systems.
Thus,
ensuring
accuracy
repeatability
model
parameter
estimation
is
a
fundamental
concern.
In
this
work,
we
structural
practical
identifiability
class
nested
compartment
pervasively
analysis
MRI
data.
We
combine
artificial
real
data
study
role
noise
estimation.
observe
that
although
all
structurally
identifiable,
strongly
depends
on
characteristics.
impact
increasing
show
how
latter
can
be
recovered
with
increased
quality.
To
complete
analysis,
results
do
not
depend
specific
tissue
characteristics
or
type
enhancement
patterns
signal.
Язык: Английский
Convergence, sampling and total order estimator effects on parameter orthogonality in global sensitivity analysis
PLoS Computational Biology,
Год журнала:
2024,
Номер
20(7), С. e1011946 - e1011946
Опубликована: Июль 17, 2024
Dynamical
system
models
typically
involve
numerous
input
parameters
whose
“effects”
and
orthogonality
need
to
be
quantified
through
sensitivity
analysis,
identify
inputs
contributing
the
greatest
uncertainty.
Whilst
prior
art
has
compared
total-order
estimators’
role
in
recovering
“true”
effects,
assessing
their
ability
recover
robust
parameter
for
use
identifiability
metrics
not
been
investigated.
In
this
paper,
we
perform:
(i)
an
assessment
using
a
different
class
of
numerical
representing
cardiovascular
system,
(ii)
wider
evaluation
sampling
methodologies
interactions
with
estimators,
(iii)
investigation
consequences
permuting
estimators
on
orthogonality,
(iv)
study
sample
convergence
resampling,
(v)
whether
positive
outcomes
are
sustained
when
model
dimensionality
increases.
Our
results
indicate
that
Jansen
or
Janon
display
efficient
minimum
uncertainty
coupled
Sobol
lattice
rule
methods,
making
them
prime
choices
calculating
influence.
This
reveals
global
analysis
is
driven.
Unconverged
indices
subject
error
therefore
true
influence
recovered.
importantly
clarifies
estimator
methodology
by
reducing
associated
ambiguities,
defining
novel
practices
modelling
life
sciences.
Язык: Английский
Structural and practical identifiability of contrast transport models for DCE-MRI
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2023,
Номер
unknown
Опубликована: Дек. 19, 2023
Abstract
Compartment
models
are
widely
used
to
quantify
blood
flow
and
transport
in
dynamic
contrast-enhanced
magnetic
resonance
imaging.
These
analyze
the
time
course
of
contrast
agent
concentration,
providing
diagnostic
prognostic
value
for
many
biological
systems.
Thus,
ensuring
accuracy
repeatability
model
parameter
estimation
is
a
fundamental
concern.
In
this
work,
we
structural
practical
identifiability
class
nested
compartment
pervasively
analysis
MRI
data.
We
combine
artificial
real
data
study
role
noise
estimation.
observe
that
although
all
structurally
identifiable,
strongly
depends
on
characteristics.
impact
increasing
show
how
latter
can
be
recovered
with
increased
quality.
To
complete
analysis,
results
do
not
depend
specific
tissue
characteristics
or
type
enhancement
patterns
signal.
Язык: Английский