Description Usage Arguments Details Value Author(s) References See Also Examples
curereg
fits parametric regression models with cure fraction for survival
data. This function extends the flexsurvreg
by the inclusion of the cure fraction
in the formulation and adds the Marshall-Olkin extreme value distribution in the comprehensive
roll of parametric distributions avaliable.
1 2 |
formula |
an object of class " |
cureformula |
a formula defining the cure rate model. In the case of no covariate effect
on the cure rate, set |
data |
the data set of class |
weights |
optional prior weights for the data. |
timedist |
survival distribution for the non-cured individuals. This can be set as:
The exponential, Weibull, log-normal and log-logistic distributions have the same
parameterization defined in |
ncausedist |
distribution of the number of competing causes of the event. This can be
set as |
subset |
optional numeric vector specifying the subset observations from the full data set. |
na.action |
a function indicating what should happen when |
inits |
optional list with the initial values for the parameters. This list should be
set as |
method |
The optimisation method to be used. |
... |
optional arguments for the For situations where the default The argument |
Note that the arguments ncausedist
and timedist
set up the model to
be fitted. This means that if ncausedist = "poisson"
and timedist = "genf"
the fitted model is obtained considering the promotion-time model with generalised F
responses. The improper density function of this model is available in dgenfpt
.
If ncausedist = "bernoulli"
and timedist = "genf"
has been set then the fitted
model is calculated considering the standard mixture model with generalised F responses.
The improper density function of this model is available in dgenfms
.
d____ms
and p____ms
correspond to the improper density and probability
functions for standard mixture models, respectively, where ____
can be any of the
distributions in timedist
. Similarly, d____pt
and p____pt
correspond
to the improper density and probability functions for promotion time models.
The relationship between the linear predictor in formula
and the time-to-event of
the non-cured elements is logarithmic as in the accelerated failure time models
(Lawless, 2003). However, for cureformula
, if ncausedist = "bernoulli"
the relationship is similar to a logistic regression model
(family = binomial(link = " logit ")
in glm
) and if
ncausedist = "poisson"
is similar to a Poisson model with logarithmic link
function (family = poisson(link = "log")
in glm
). See the references for
more details.
A list of class "curereg" containing information about the fitted model. Components of interest to users may include:
call |
the matched call. |
coefficients |
a named vector of coefficients obtained via Maximum Likelihood (see Details). |
std.error |
a named vector of the estimated standard errors for the coefficients (see Details). |
vcov |
A matrix of the estimated covariances between the coefficient estimates in the predictor of the model. |
loglik |
log-likelihood. |
AIC |
AIC the (generalized) Akaike Information Criterion for the fitted model. |
Rumenick Pereira da Silva rumenickps@gmail.com
Jackson, C.H. Christopher Jackson (2015). flexsurv: Flexible Parametric Survival and Multi-State Models. R package version 0.7. https://CRAN.R-project.org/package=flexsurv.
Maller, R. A., & Zhou, X. (1996). Survival analysis with long-term survivors. New York: Wiley.
Lawless, J. F. (2011). Statistical models and methods for lifetime data (Vol. 362). John Wiley & Sons.
Ortega, E. M., Cancho, V. G., & Paula, G. A. (2009). Generalized log-gamma regression models with cure fraction. Lifetime Data Analysis, 15(1), 79-106.
Peng, Y., Dear, K. B., & Denham, J. W. (1998). A generalized F mixture model for cure rate estimation. Statistics in medicine, 17(8), 813-830.
confint.curereg
for confidence intervals for the coefficients,
curefraction
for predict cure fractions from fitted curereg
model,
plot.curereg
and lines.curereg
to plot fitted survival, hazards
and cumulative hazards from models fitted by curereg
.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 | ## fit Marshall-Olkin extended extreme value standart mixture model
data(e1684)
fitmo <- curereg(Surv(FAILTIME, FAILCENS) ~ TRT + SEX + AGE, cureformula = ~ TRT + SEX + AGE,
data = e1684, timedist = "moeev", ncausedist = "bernoulli")
# Output of 'curereg' object
fitmo
# Extract Model Coefficients
coef(fitmo)
# Extract model coefficients:
# Terms: failure time distribution model
coef(fitmo, terms = "time")
# Terms: cure probability model
coef(fitmo, terms = "cure")
# Object summaries
summary(fitmo)
# Information criterion
AIC(fitmo) # Akaike information criterion
BIC(fitmo) # Bayesian information criterion
# Extract Log-Likelihood
logLik(fitmo)
# Calculate Variance-Covariance Matrix for a Fitted Model Object
vcov(fitmo)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.