Description Usage Arguments Details Value References Examples
This is the main function to obtain the point estimator and pointwise confidence interval for the cumulative incidence function.
1 2 |
formulas |
A list of length K contains formula objects, where K is the number of failures. Each element is a formula object, with the response on the left of a ~ operator, and the predictors on the right. The response must be a survival object as returned by the |
data |
A data frame with n observations and p number of covariates. |
group |
The name of the group covariates (if any). If specified, the cumulative incidence function will be estimated for each group separately. Default is |
group.in.train |
This argument is valid only when the group argument is specified. If group is presented in both data and newdata, use |
newdata |
A data frame or a matrix used for prediction. If not specified, the original data will be used instead. |
compute.CI |
A logic operator for whether to compute the pointwise confidence interval. The default is |
alpha |
The significance level. The default is 0.05, which computes the |
transform |
An indicator if transformation is used for the confidence interval. Only log-log transformation is implemented currently. |
save |
An option to save the computed S0 and S1. It is highly recommended for large-scale dataset to improve the computational efficiency. Default is |
This is the main function to compute the point estimator and pointwise confidence interval for the cumulative incidence function. We implemented a flexible function to allow prediction of cumulative incidence function under the cause-specific proportion hazards model in scenarios: 1) when group segmentation is not specified in training and test data; 2) when group segmentation is specified in both training and test data; 3) group segmentation is presented in the test data only; 4) the type of failure is associated with a different setts of covariates. To predict, user can input the new values of all covariates as a vector. If the newdata
is specified, the mean of all observations with corresponding covariates will be used in the prediction. In addition, the generic functions such as plot()
, summary()
are available for the newly defined CIF
class.
When group=NULL
, est.CIF
returns a list of length K. For each sublist, it contains:
model.fit |
Details of model fit, such as parameter estimates, standard error, etc. |
call |
The survival object used for a the k^th type of failure. |
Estimate |
A data frame contains events.time, overall survival, cumulative incidence, upper and lower bounds (if |
When group
is specified, est.CIF
returns a list of length that equals to the number of levels in group
. For each sublist, it contains a list of length K.
Cheng, S. C., Jason P. Fine, and L. J. Wei. "Prediction of cumulative incidence function under the proportional hazards model." Biometrics (1998): 219-228.
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 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | ## Not run:
data("train.data")
data("test.data2")
#1.Example: no group in training and test #
formulas <- list(as.formula("Surv(dtime, prostate) ~ AG + WT + PF + HX +HG + SZ + SG + RX"),
as.formula("Surv(dtime, cardio) ~ AG + WT + PF + HX +HG + SZ + SG + RX"),
as.formula("Surv(dtime, Other_causes) ~ AG + WT + PF + HX +HG + SZ + SG + RX"))
CIF.output <- est.CIF(formulas, data = train.data, newdata = test.data2,
group=NULL, compute.CI=T,save=F)
plot(CIF.output)
#2. Example: group in training data and test data #
formulas <- list(as.formula("Surv(dtime, prostate) ~ AG + WT + PF + HX +HG + SZ + SG + RX"),
as.formula("Surv(dtime, cardio) ~ AG + WT + PF + HX +HG + SZ + SG + RX"),
as.formula("Surv(dtime, Other_causes) ~ AG + WT + PF + HX +HG + SZ + SG + RX"))
CIF.output <- est.CIF(formulas, data=train.data, group="RX", group.in.train=T,
newdata=test.data2, compute.CI=T, alpha=0.05, transform="log-log", save=F)
plot(CIF.output, group="RX")
#3. Example: group not in training data and but in test data #
formulas <- list(as.formula("Surv(dtime, prostate) ~ AG + WT + PF + HX +HG + SZ + SG"),
as.formula("Surv(dtime, cardio) ~ AG + WT + PF + HX +HG + SZ + SG "),
as.formula("Surv(dtime, Other_causes) ~ AG + WT + PF + HX +HG + SZ + SG"))
CIF.output <- est.CIF(formulas, data=train.data, group="RX", group.in.train=F, newdata=test.data2,
compute.CI=T, alpha=0.05, transform="log-log", save=F)
plot(CIF.output, group="RX")
#4. Example: 1)group not in training data and but in test data
# 2)different risk factors associated with cause of failure #
formulas <- list(as.formula("Surv(dtime, prostate) ~ AG + WT + PF + HX +HG"),
as.formula("Surv(dtime, cardio) ~ PF + HX +HG "),
as.formula("Surv(dtime, Other_causes) ~ AG + WT + PF + HX +HG + SZ"))
CIF.output <- est.CIF(formulas, data=train.data, group="RX", group.in.train=F,
newdata=test.data2,compute.CI=T, alpha=0.05, transform="log-log", save=T)
plot(CIF.output, group="RX")
#5. Example: different risk factors associated with cause of failure #
formulas <- list(as.formula("Surv(dtime, prostate) ~ AG + WT + PF + HX +HG"),
as.formula("Surv(dtime, cardio) ~ PF + HX +HG + SZ"),
as.formula("Surv(dtime, Other_causes) ~ AG + WT + PF + HX +HG + SZ + SG"))
CIF.output <- est.CIF(formulas, data=train.data, group=NULL, group.in.train=F, newdata=test.data2,
compute.CI=T, alpha=0.05, transform="log-log", save=F)
plot(CIF.output)
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.