addhazard (version 1.0.0)

predict.ah.2ph: Prediction Based on the Additive Hazards Model Fitted from Two-phase Sampling

Description

This function predicts a subject's overall hazard rates at given time points based on this subject's covariate values. The prediction function is an object from ah.2ph. The estimating procedures follow Hu (2014).

Usage

"predict"(object, newdata, newtime, ...)

Arguments

object

an object of class inhering from "ah.2ph".

newdata

a dataframe of an individual's predictors.

newtime

a given sequence of time points at which the prediction is performed.

...

further arguments passed to or from other methods.

Value

A dataframe including the given time points, predicted hazards, their standard errors, their variances, the phase I component of the variance for predicted hazards and the phase II component of the variance.

References

Jie Hu (2014) A Z-estimation System for Two-phase Sampling with Applications to Additive Hazards Models and Epidemiologic Studies. Dissertation, University of Washington.

See Also

ah.2ph for fitting the additive hazards model with two-phase sampling and nwtsco for the description of nwtsco dataset

Examples

library(survival)
### load data
nwts <- nwtsco[1:100,]

### create strata based on  institutional histology and disease status
nwts$strt <- 1+nwts$instit
### add a stratum containing all (relapsed) cases
nwts$strt[nwts$relaps==1] <- 3

### assign phase II subsampling probabilities
### oversample unfavorable histology (instit =1) and cases
### Pi = 0.5 for instit =0, Pi =1 for instit =1 and relaps =1
nwts$Pi<-  0.5 * (nwts$strt == 1) + 1 * (nwts$strt == 2) + 1 * (nwts$strt == 3)

### generate phase II sampling indicators
N <- dim(nwts)[1]
nwts$in.ph2 <-  rbinom(N, 1, nwts$Pi)

### fit an additive hazards model to  two-phase sampling data without calibration
fit1 <- ah.2ph(Surv(trel,relaps) ~ age + histol, data = nwts, R = in.ph2, Pi = Pi, robust = FALSE)



###  input the new data for prediction
newdata <- nwtsco[101,]
###  based on the fitted model fit1, perform prediction at time points t =3 and t= 5
predict(fit1, newdata, newtime = c(3,5))

### fit an additve hazards model to  two-phase sampling data with calibration
### The calibration variable is stage
fit2 <- ah.2ph(Surv(trel,relaps) ~ age + histol, data = nwts, R = in.ph2, Pi = Pi,
                                   robust = FALSE, calibration.variables = stage)

### based on the fitted model fit2, perform prediction at time points t =3 and t= 5
predict(fit2, newdata, newtime = c(3,5))