Statistical Specifications
Source:vignettes/statistical-specification.Rmd
statistical-specification.Rmd
IMPORTANT
Please note that this document is currently a work-in-progress and does not contain complete information for this package yet.
Survival Model Specification
This package can only be used to fit proportional hazards models of the form:
Where:
- is a parametric baseline hazard function
- is the event time
- is a vector of parameters that parameterise the baseline hazard
- is an arbitrary vector of parameters for subject specified by the longitudinal model
- is a link function that maps to a contribution to the log-hazard function where simply indexes the given link function
- is the subjects covariate design matrix
- is the corresponding coefficients to scale the design matrix covariates contribution to the log-hazard function
The following sections outline the available distributions to users which can be selected for the baseline hazard . Please note that some of these distributions do not have the proportional-hazards property meaning that the resulting survival model corresponding to the hazard will not be of the same parametric family as the baseline distribution with the hazard .
Longitudinal Model Specification
Random-Slope Model
where:
- is the tumour size for subject at timepoint
- is the intercept for subject
- is the random slope for subject with
- is the mean for the random slope within treatment arm
- is the variance term for the slopes
- is the error term with
- is the treatment arm index for subject
- is the study index for subject
Derivative of the SLD Trajectory Link / Growth Parameter Link
Accessible via linkDSLD()
&
linkGrowth()
Identity Link
Accessible via linkIdentity()
Stein-Fojo Model
Where:
- is the subject index
- is the visit index
- is the observed tumour measurements
- is the expected sum of longest diameter for subject at time point
- is the time since first treatment for subject at visit
- is the subject baseline SLD measurement
- is the subject kinetics shrinkage parameter
- is the subject kinetics tumour growth parameter
- is the subject proportion of cells affected by the treatment
- is the treatment arm index for subject
- is the study index for subject
- is the population mean for parameter in group
- is the population variance for parameter .
If using the non-centred parameterisation then the following alternative formulation is used:
Where:
- is a random effects offset on parameter for subject
If using the unscaled variance parameterisation then the following alternative formulation is used: $$ y_{ij} &\sim N(SLD_{ij},\ \sigma^2) $$
Derivative of the SLD Trajectory Link
Accessible via linkDSLD()
Time to Growth Link
Accessible via linkTTG()
Identity Link
Accessible via linkIdentity()
Growth Parameter Link
Accessible via linkGrowth()
Shrinkage Parameter Link
Accessible via linkShrinkage()
Population Quantities
Note that when generating population quantities for the , and parameters the median of the distribution is used. This is because this has the same interpretation as using a non-centred parameterisation and setting the “random effects” term to be 0. For the parameter the mean of the distribution is used.
Generalized Stein-Fojo (GSF) Model
Where:
- is the subject index
- is the visit index
- is the observed tumour measurements
- is the expected sum of longest diameter for subject at time point
- is the time since first treatment for subject at visit
- is the subject baseline SLD measurement
- is the subject kinetics shrinkage parameter
- is the subject kinetics tumour growth parameter
- is the subject proportion of cells affected by the treatment
- is the treatment arm index for subject
- is the study index for subject
- is the population mean for parameter in group
- is the population variance for parameter in group .
If using the non-centred parameterisation then the following alternative formulation is used:
Where:
- is a random effects offset on parameter for subject
If using the unscaled variance parameterisation then the following alternative formulation is used: $$ y_{ij} &\sim N(SLD_{ij},\ \sigma^2) $$
Derivative of the SLD Trajectory Link
Accessible via linkDSLD()
Time to Growth Link
Accessible via linkTTG()
Identity Link
Accessible via linkIdentity()
Growth Parameter Link
Accessible via linkGrowth()
Shrinkage Parameter Link
Accessible via linkShrinkage()
Claret-Bruno Model
Where:
- is the subject index
- is the visit index
- is the observed tumour measurements
- is the expected sum of longest diameter for subject at time point
- is the time since first treatment for subject at visit
- is the subject baseline SLD value.
- is the subject tumour growth rate.
- is the subject treatment resistance rate.
- is the subject treatment growth inhibition response.
- is the treatment arm index for subject
- is the study index for subject
- is the population mean for parameter in group
- is the population variance for parameter in group
If using the non-centred parameterisation then the following alternative formulation is used:
If using the unscaled variance parameterisation then the following alternative formulation is used: $$ y_{ij} &\sim N(SLD_{ij},\ \sigma^2) $$
Derivative of the SLD Trajectory Link
Accessible via linkDSLD()
Time to Growth Link
Accessible via linkTTG()
Identity Link
Accessible via linkIdentity()
Growth Parameter Link
Accessible via linkGrowth()
Post-Processing
Brier Score
The Brier Score is used to measure a models predictive performance.
In the case of Survival Models it is the weighted squared difference
between whether a subject died within a given time interval and the
models estimated probability of them dying within said interval. Within
jmpost
the following formula (as described in Blanche et al. (2015)) has been implemented:
Where:
- is the observed event/censor time for subject
- is the event indicator which is 1 if is an event and 0 if is censored
- is the indicator which is 1 if is true else 0
- is the event indicator function for subject e.g.
- is a model predicted probability of subject dying before time
- is the Kaplan-Meier estimator of survival function of the censoring time at
Note that by default
is estimated by
and that in the case of ties event times are always considered to have
occurred before censored times (this is in contrast to
survival::survfit
which regards censored times as coming
before event times when estimating the censoring distribution). Both of
these default options can be changed if required.