survivROC: Generates the ROC curve at a given time point given the...

Description Usage Arguments Details Value Note Author(s) References See Also Examples

View source: R/survivROC.R

Description

The function generates the Receiver-Operator Curve (ROC) at the specified time point using predicted and observed data in the presence of censored subjects. It so plots (1 - specificity) against the (specificity) at the designated cut off points. It is based on Patrick Heagerty's survivalROC function in the survivalROC package.

Usage

1
survivROC(Stime, status, marker, entry = NULL, predict.time, cut.values = NULL, plot = TRUE)

Arguments

Stime

The observed survival times of the patients.

status

The censoring status of the patient. 1 for a censored patient, and 0 for a patient who has an event.

marker

The predicted survival time of the patients.

entry

The time of entry of the patients, set to NULL by default.

predict.time

The time point for which the ROC curve is to be plotted.

cut.values

The cut off values for which the ROC curves are to be constructed.

plot

A logical argument that specifies whether a plot is to be generated (TRUE) or not (FALSE). The argument is set to TRUE by default.

Details

This function is basically the survivROC function in Patrick Heagerty's survivROC package, with slight modifications to it to better suit our purpose. Unlike Heagerty's function it only performs the calculations using the KM estimator and does not provide any other methods as options.

Value

cut.values

the cut off values for which the sensitivity and (1-specificity) were calculated

Comp.Specificity

(1 - Specificity) as was calculated for every cut off value.

Sensitivity

The Sensitivity as was calculated for every cut off value

predict.time

The time point for which the ROC curve is calculated

Survival

The value of the estimated survivor function (using the KM estimator) at "predict.time"

AUC

The value of the area under the estimated ROC curve

Note

It is important to note that using the Kaplan-Meier estimator as the method for estimating the survival function results in non-monotonous curves in some instances.

Author(s)

Douaa Mugahid

References

Heagerty,P., Lumely T. & Pepe M.(2000). Time-dependent ROC curves for censored survival data & a diagnostic marker. Biometrics, 56(2), 337-344.

See Also

survivAURC

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
True_STs <- c(1.416667,2.75,2.416667,2.583333,2.166667,2.5,2.5,1.833333,1.25,0.6666667,1,6.583333,6.5,6.666667,2.75,1.666667,1.166667,2.833333,3.583333,6.166667,6.166667,
3.416667,6.083333,1.833333,5.583333,0.75,5.75,5.5,0.5833333,7.666667,5,2.833333,1.333333,5.083333,0.8333333,1.5,4.75,3.416667,4.666667,1.916667,4.666667,7.416667,0.9166667,
1.083333,3.75,3.25,3,2.416667,2.75,2.5,2.666667,4.5,4.416667,1.5,0.8333333,3.166667,3.833333,3.833333,0.4166667,3.333333,2.75,3.083333,0.3333333,0.25,0.6666667,1.833333,
2.333333,3.416667,3.416667,3,0.6666667,0.75,2.166667,1,1.416667,1.333333,1.166667,1.166667,0.4166667,1.25,1.166667,1.083333)
Predicted_STs <- c(6.030591,6.014457,3.545584,5.414229,6.41576,9.393992,5.542331,6.890859,8.090213,4.98545,2.77357,6.275699,9.163978,7.511511,9.531218,7.63715,10.08977,
11.12364,3.982502,5.441881,12.61404,12.21851,17.05850,12.78141,16.22795,21.48544,6.281354,13.83925,8.859929,6.104142,8.255909,2.335526,6.564962,2.335761,9.33772,12.62540,
10.97276,15.63089,8.01967,5.817267,5.59897,4.340784,32.40319,33.74123,27.45024,26.31024,26.88833,24.34707,32.06541,38.90473,17.37102,15.11059,8.772035,14.24816,7.852889,
7.79996,5.601459,2.802408,35.77047,24.34717,30.65796,25.93927,20.64544,22.04807,19.15037,23.83430,1.876557,3.937208,6.526354,5.886377,9.301074,12.4657,14.49783,15.41502,
2.860931,2.541947,4.543111,4.525553,4.148272,3.986912,6.246755,6.89523)
censored <- c(0,1,1,1,1,0,1,1,0,1,0,1,1,1,1,0,0,0,0,1,1,1,1,0,1,0,1,1,0,1,1,0,0,1,0,0,1,0,1,0,1,1,0,0,1,1,1,0,1,1,1,1,1,0,1,1,1,1,0,1,1,1,1,0,0,0,1,1,1,1,0,0,1,0,1,1,1,1,1,1,1,
1)
survivROC(Stime=True_STs,status=censored, marker=Predicted_STs,predict.time=5)

RCASPAR documentation built on Nov. 17, 2017, 9:27 a.m.