/
casewise.R
456 lines (408 loc) · 16.7 KB
/
casewise.R
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
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
##' Estimates the casewise concordance based on Concordance and marginal estimate using timereg and performs test for independence
##'
##' @title Estimates the casewise concordance based on Concordance and marginal estimate using timereg and performs test for independence
##' @details Uses cluster based conservative standard errors for marginal and sometimes only the uncertainty of the concordance estimates. This works prettey well, alternatively one can use also the
##' funcions Casewise for a specific time point
##' @param conc Concordance
##' @param marg Marginal estimate
##' @param test Type of test for independence assumption. "conc" makes test on concordance scale and "case" means a test on the casewise concordance
##' @param p check that marginal probability is greater at some point than p
##' @author Thomas Scheike
##' @aliases casewise.test slope.process casewise.bin
##' @examples
##' \donttest{ ## Reduce Ex.Timings
##' library("timereg")
##' data("prt",package="mets");
##' prt <- force.same.cens(prt,cause="status")
##'
##' prt <- prt[which(prt$id %in% sample(unique(prt$id),7500)),]
##' ### marginal cumulative incidence of prostate cancer
##' times <- seq(60,100,by=2)
##' outm <- timereg::comp.risk(Event(time,status)~+1,data=prt,cause=2,times=times)
##'
##' cifmz <- predict(outm,X=1,uniform=0,resample.iid=1)
##' cifdz <- predict(outm,X=1,uniform=0,resample.iid=1)
##'
##' ### concordance for MZ and DZ twins
##' cc <- bicomprisk(Event(time,status)~strata(zyg)+id(id),
##' data=prt,cause=c(2,2))
##' cdz <- cc$model$"DZ"
##' cmz <- cc$model$"MZ"
##'
##' ### To compute casewise cluster argument must be passed on,
##' ### here with a max of 100 to limit comp-time
##' outm <- timereg::comp.risk(Event(time,status)~+1,data=prt,
##' cause=2,times=times,max.clust=100)
##' cifmz <- predict(outm,X=1,uniform=0,resample.iid=1)
##' cc <- bicomprisk(Event(time,status)~strata(zyg)+id(id),data=prt,
##' cause=c(2,2),se.clusters=outm$clusters)
##' cdz <- cc$model$"DZ"
##' cmz <- cc$model$"MZ"
##'
##' cdz <- casewise.test(cdz,cifmz,test="case") ## test based on casewise
##' cmz <- casewise.test(cmz,cifmz,test="conc") ## based on concordance
##'
##' plot(cmz,ylim=c(0,0.7),xlim=c(60,100))
##' par(new=TRUE)
##' plot(cdz,ylim=c(0,0.7),xlim=c(60,100))
##'
##' slope.process(cdz$casewise[,1],cdz$casewise[,2],iid=cdz$casewise.iid)
##'
##' slope.process(cmz$casewise[,1],cmz$casewise[,2],iid=cmz$casewise.iid)
##'
##' }
##' @export
casewise.test <- function(conc,marg,test="no-test",p=0.01)
{ ## {{{
if (sum(marg$P1>p)==0) stop("No timepoints where marginal > ",p,"\n");
time1 <- conc$time; time2 <- marg$time[marg$P1>0.01]
mintime <- max(time1[1],time2[1])
maxtime <- min(max(time1),max(time2))
timer <- seq(mintime, maxtime,length=100)
dtimer <- timer[2]-timer[1]
margtime <- cpred(cbind(marg$time,c(marg$P1)),timer)[,2]
concP1 <- cpred(cbind(conc$time,c(conc$P1)),timer)[,2]
se.margtime <- cpred(cbind(marg$time,c(marg$se.P1)),timer)
se.concP1 <- cpred(cbind(conc$time,c(conc$se.P1)),timer)
outtest <- NULL
casewise.iid <- NULL
casewise <- concP1/margtime
se.casewise2 <- as.matrix(se.concP1[,2]/margtime,ncol=100)
se.margtime <- as.matrix(se.margtime[,2],ncol=100)
se.casewise <- NULL;
outtest <- NULL;
if (is.null(conc$P1.iid) || is.null(marg$P1.iid)) {
cat("Warning, need iid decomposition for correct se's for Concordance \n");
} else {
conciid <- cpred(cbind(conc$time,conc$P1.iid),timer)[,-1]
nconc <- colnames(conc$P1.iid)
P1iid <- cpred(cbind(marg$time,marg$P1.iid),timer)[,-1]
P1iid2 <- 2*P1iid*margtime;
se.p2 <- apply(P1iid2^2,1,sum)^.5
conciid <- (P1iid2[,nconc]-conciid) ### iid of conc-p1^2 test
c.iid <- casewise.iid <- conciid/margtime - P1iid[,nconc]*casewise
se.casewise <- apply(casewise.iid^2,1,sum)^.5
dif.casewise.iid <- conciid/margtime ### iid of casewise test
if (test=="case")
{
diff.iidcase <- apply(dif.casewise.iid,2,sum)*dtimer;
sd.pepem <- sum(diff.iidcase^2)^.5
diff.pepem <- sum((casewise-margtime))*dtimer
z.pepem <- diff.pepem/sd.pepem
pval.pepem <- 2*pnorm(-abs(z.pepem))
outtest <- cbind(diff.pepem,sd.pepem,z.pepem,pval.pepem)
### test for constant casewise concordance
diff.const <- (casewise-mean(casewise))
iid.constant <- (c.iid - matrix(apply(c.iid,2,mean),nrow(c.iid),ncol(c.iid),byrow=T))
se.const <- apply(iid.constant^2,1,sum)^.5
test.constant <- max(abs(diff.const/se.const))
sim.maxs <- apply(abs(iid.constant/se.const),1,max)
pval.const <- timereg::pval(sim.maxs,test.constant)
outtest <- cbind(diff.pepem,sd.pepem,z.pepem,pval.pepem,test.constant,pval.const)
colnames(outtest) <- c("cum dif.","sd","z","pval","constant-case","pval")
rownames(outtest) <- "pepe-mori"
} else if (test=="conc") {
diff.iid <- apply(conciid,2,sum)*dtimer
sd.pepem <- sum(diff.iid^2)^.5
diff.pepem <- sum((concP1-margtime^2))*dtimer
z.pepem <- diff.pepem/sd.pepem
pval.pepem <- 2*pnorm(-abs(z.pepem))
outtest <- cbind(diff.pepem,sd.pepem,z.pepem,pval.pepem)
colnames(outtest) <- c("cum dif.","sd","z","pval")
rownames(outtest) <- "pepe-mori"
}
}
concout <- cbind(timer,concP1,se.concP1[,2])
colnames(concout) <- c("time","concordance","se")
margout <- cbind(timer,margtime,se.margtime)
colnames(margout) <- c("time","cif","se")
casewiseout <- cbind(timer,casewise,se.casewise,se.casewise2)
colnames(casewiseout) <- c("time","casewise","se","se2")
difout <- cbind(timer,concP1-margtime^2,apply(conciid^2,1,sum)^.5)
out <- list(casewise=casewiseout,marg=margout,conc=concout,casewise.iid=casewise.iid,
test=outtest,mintime=mintime,maxtime=maxtime,same.cluster=TRUE,testtype=test)
class(out) <- "casewise"
return(out)
} ## }}}
##' @export
slope.process <- function(time,y,iid=NULL)
{ ## {{{
ctime <- scale(time)
caselm <- lm(y ~ ctime)
slope <- coef(caselm)[2]
if (!is.null(iid)) {
diff.iid <- iid
lm.iid <- c()
for (i in 1:ncol(iid))
{
lmiid <- lm(iid[,i] ~ ctime)
lm.iid <- rbind(lm.iid,coef(lmiid))
}
se.slope <- apply(lm.iid,2,sd)^.5
} else {se.slop <- NULL}
z.slope <- slope/se.slope[2]
pval <- 2*(1-pnorm(abs(z.slope)))
out <- list(intercept=coef(caselm)[1],slope=slope,se.slope=se.slope,pval.slope=pval)
return(out)
} ## }}}
##' .. content for description (no empty lines) ..
##'
##' @title Estimates the casewise concordance based on Concordance and marginal estimate using prodlim but no testing
##' @param conc Concordance
##' @param marg Marginal estimate
##' @param cause.marg specififes which cause that should be used for marginal cif based on prodlim
##' @author Thomas Scheike
##' @examples
##' \donttest{ ## Reduce Ex.Timings
##' library(prodlim)
##' data(prt);
##' prt <- force.same.cens(prt,cause="status")
##'
##' ### marginal cumulative incidence of prostate cancer##'
##' outm <- prodlim(Hist(time,status)~+1,data=prt)
##'
##' times <- 60:100
##' cifmz <- predict(outm,cause=2,time=times,newdata=data.frame(zyg="MZ")) ## cause is 2 (second cause)
##' cifdz <- predict(outm,cause=2,time=times,newdata=data.frame(zyg="DZ"))
##'
##' ### concordance for MZ and DZ twins
##' cc <- bicomprisk(Event(time,status)~strata(zyg)+id(id),data=prt,cause=c(2,2),prodlim=TRUE)
##' cdz <- cc$model$"DZ"
##' cmz <- cc$model$"MZ"
##'
##' cdz <- casewise(cdz,outm,cause.marg=2)
##' cmz <- casewise(cmz,outm,cause.marg=2)
##'
##' plot(cmz,ci=NULL,ylim=c(0,0.5),xlim=c(60,100),legend=TRUE,col=c(3,2,1))
##' par(new=TRUE)
##' plot(cdz,ci=NULL,ylim=c(0,0.5),xlim=c(60,100),legend=TRUE)
##' summary(cdz)
##' summary(cmz)
##' }
##' @export
casewise <- function(conc,marg,cause.marg)
{ ## {{{
if (missing(cause.marg)) stop("Please specify cause of marginal (as given in Event object)")
if ((!inherits(conc,"prodlim")) || (!inherits(marg,"prodlim"))) stop("Assumes that both models are based on prodlim function \n");
time1 <- conc$time
time2 <- marg$time
cause.prodlim <- match(as.character(cause.marg),levels(prodlim::getEvent(marg$model.response)))
if (is.na(cause.prodlim)) stop("Cause did not match marginal model")
mintime <- max(time1[1],time2[1])
maxtime <- min(max(time1),max(time2))
timer <- seq(mintime, maxtime,length=100)
dtimer <- timer[2]-timer[1]
out <- conc
out$time <- timer
if (inherits(marg,"comp.risk")) margtime <- cpred(cbind(marg$time,c(marg$P1)),timer)[,2] else if (inherits(marg,"prodlim")) {
cuminc <- data.frame(marg$cuminc)[,cause.prodlim];
se.cuminc <- data.frame(marg$se.cuminc)[,cause.prodlim];
margtime <- cpred(cbind(marg$time,c(cuminc)),timer)[,2];
se.margtime <- cpred(cbind(marg$time,c(se.cuminc)),timer)[,2];
} else stop("marginal cumulative incidence comp.risk or prodlim output\n");
if (inherits(conc,"comprisk")) concP1 <- cpred(cbind(conc$time,c(conc$P1)),timer)[,2]
else if (inherits(conc,"prodlim")) {
conc.cuminc <- data.frame(conc$cuminc)[,1]
conc.se.cuminc <- data.frame(conc$se.cuminc)[,1]
se.P1 <- cpred(cbind(conc$time,conc.se.cuminc),timer)[,2]
concP1 <- cpred(cbind(conc$time,conc.cuminc),timer)[,2]
}
concordance<- cbind(timer,concP1,se.P1)
colnames(concordance) <- c("time","cif","se.cif")
P1 <- concP1/margtime
se.P1 <- se.P1/margtime
med <- (margtime>0) & (concP1 > 0)
out$P1 <- P1[med]
out$se.P1 <- se.P1[med]
out$timer <- timer[med]
margout <- cbind(timer,margtime,se.margtime)
colnames(margout) <- c("time","cif","se.cif")
probout <- cbind(out$timer,out$P1,out$se.P1)
colnames(probout) <- c("time","casewise conc","se casewise")
out <- list(casewise=probout,marg=margout,concordance=concordance,test=NULL)
class(out) <- "casewise"
return(out)
} ## }}}
##' @export
casewise.bin <- function(nc,nd)
{ ## {{{
ud <- glm(cbind(nc,round(0.5*nd+nc))~ +1,family=binomial())
udci <- confint(ud)
pud <- predict(ud,se.fit=TRUE,type="response")
return(list(p.casewise=pud$fit,ci.casewise=exp(udci)))
} ## }}}
##' @export
plot.casewise <- function(x,ci=NULL,lty=NULL,ylim=NULL,col=NULL,xlab="time",ylab="concordance",
legend=FALSE,add=FALSE, ...)
{ ## {{{
if (is.null(col)) col <- 1:3
if (is.null(lty)) lty <- 1:3
if (is.null(ylim)) ylim=range(c(x$casewise[,2],x$marg[,2]))
if (add) {
lines(x$casewise[,1],x$casewise[,2],type="s",lty=lty[1],col=col[1],...)
} else {
plot(x$casewise[,1],x$casewise[,2],type="s",ylim=ylim,lty=lty[1],col=col[1],xlab=xlab,ylab=ylab,...)
}
if (!is.null(ci)) {
ul <- x$casewise[,2]+qnorm(1-(1-ci)/2)* x$casewise[,3]
nl <- x$casewise[,2]-qnorm(1-(1-ci)/2)* x$casewise[,3]
lines(x$casewise[,1],ul,type="s",ylim=ylim,lty=lty[3],col=col[3])
lines(x$casewise[,1],nl,type="s",ylim=ylim,lty=lty[3],col=col[3])
}
lines(x$marg[,1],x$marg[,2],lty=lty[2],col=col[2],type="s")
if (legend==TRUE) legend("topleft",lty=lty[1:2],col=col[1:2],c("Casewise concordance","Marginal estimate"))
} ## }}}
##' @export
summary.casewise <- function(object,marg=FALSE,...)
{ ## {{{
cat("Casewise concordance and standard errors \n");
print(signif(cbind(object$casewise),3))
cat("\n");
if (marg==TRUE) {
cat("Marginal cumulative incidence and standard errors \n");
print(signif(cbind(object$marg),3))
}
print(object,...)
} ## }}}
##' prints Concordance test
##'
##' @title prints Concordance test
##' @param x output from casewise.test
##' @param digits number of digits
##' @param \dots Additional arguments to lower level functions
##' @author Thomas Scheike
##' @export
print.casewise <- function(x,digits=3,...)
{ ## {{{
cat("\n")
if (!is.null(x$test)) {
cat("Pepe-Mori type test for H_0: conc_1(t)= conc_2(t)\n")
if (x$same.cluster==TRUE) cat("Assuming same clusters for the two functions\n") else
cat("Assuming independence for estimators\n");
cat(paste("Time.range =",signif(x$mintime,3),"--",signif(x$maxtime,3),"\n\n"));
prmatrix(signif(x$test,digits))
invisible(x)
}
} ## }}}
##' .. content for description (no empty lines) ..
##'
##' @title Concordance test Compares two concordance estimates
##' @param conc1 Concordance estimate of group 1
##' @param conc2 Concordance estimate of group 2
##' @param same.cluster if FALSE then groups are independent, otherwise estimates are based on same data.
##' @author Thomas Scheike
##' @export
test.conc <- function(conc1,conc2,same.cluster=FALSE)
{ ## {{{
time <- time1 <- conc1$time
time2 <- conc2$time
mintime <- max(time1[1],time2[1])
maxtime <- min(max(time1),max(time2))
timer <- seq(mintime, maxtime,length=100)
dtimer <- timer[2]-timer[1]
conc2timer <- cpred(cbind(conc2$time,c(conc2$P1)),timer)[,2]
conc1timer <- cpred(cbind(conc1$time,c(conc1$P1)),timer)[,2]
outtest <- NULL
if (is.null(conc1$P1.iid) || is.null(conc2$P1.iid))
stop("Must give iid represenation for both estimators\n");
if (!is.null(conc1$P1.iid) && !is.null(conc2$P1.iid)) {
if ( ((ncol(conc1$P1.iid[,])-ncol(conc2$P1.iid[,]))!=0) && same.cluster==TRUE)
cat("Warning, not same number of iid residuals for concordance and marginal estimate\n");
}
if (!is.null(conc1$P1.iid)) if (!is.null(conc2$P1.iid)) {
### iid version af integraler
conc2P1.iid <- cpred(cbind(conc2$time,conc2$P1.iid[,]),timer)[,-1]
conc1P1.iid <- cpred(cbind(conc1$time,conc1$P1.iid[,]),timer)[,-1]
if ( (ncol(conc1$P1.iid)==ncol(conc2$P1.iid)) && same.cluster==TRUE) {
diff.iid <- conc1P1.iid-conc2P1.iid
sdiff.iid <- apply(diff.iid,2,sum)*dtimer
sd.pepem <- sum(sdiff.iid^2)^.5
} else {
diff2.iid <- conc2P1.iid
sdiff2.iid <- apply(diff2.iid,2,sum)*dtimer
var2.pepem <- sum(sdiff2.iid^2)
diff1.iid <- conc1P1.iid
sdiff1.iid <- apply(diff1.iid,2,sum)*dtimer
var1.pepem <- sum(sdiff1.iid^2)
sd.pepem <- (var1.pepem+var2.pepem)^.5
}
diff.pepem <- sum(conc2timer-conc1timer)*dtimer
### print(cbind(conc2timer,conc1timer))
z.pepem <- diff.pepem/sd.pepem
pval.pepem <- 2*pnorm(-abs(z.pepem))
outtest <- cbind(diff.pepem,sd.pepem,z.pepem,pval.pepem)
colnames(outtest) <- c("cum dif.","sd","z","pval")
rownames(outtest) <- "pepe-mori"
### print(outtest,4)
### prmatrix(outtest,3)
}
outtest <- list(test=outtest,mintime=mintime,maxtime=maxtime,same.cluster=same.cluster)
###attr(out,"class") <- rev(attr(out,"class"))
class(outtest) <- "testconc"
return(outtest)
} ## }}}
##' convert to timereg object
##'
##' @title Convert to timereg object
##' @param obj no use
##' @author Thomas Scheike
##' @export
back2timereg <- function(obj)
{ ## {{{
out <- obj
attr(out,"class") <- rev(attr(out,"class"))
return(out)
} ## }}}
##' Estimates the casewise concordance based on Concordance and marginal estimate using binreg
##'
##' @title Estimates the casewise concordance based on Concordance and marginal estimate using binreg
##' @details Uses cluster iid for the two binomial-regression estimates standard errors better than those of casewise that are often conservative.
##' @param concbreg Concordance
##' @param margbreg Marginal estimate
##' @param zygs order of zygosity for estimation of concordance and casewise.
##' @param newdata to give instead of zygs.
##' @param ... to pass to estimate function
##' @author Thomas Scheike
##' @examples
##' data(prt)
##' prt <- force.same.cens(prt,cause="status")
##'
##' dd <- bicompriskData(Event(time, status)~strata(zyg)+id(id), data=prt, cause=c(2, 2))
##' newdata <- data.frame(zyg=c("DZ","MZ"),id=1)
##'
##' ## concordance
##' bcif1 <- binreg(Event(time,status)~-1+factor(zyg)+cluster(id), data=dd,
##' time=80, cause=1, cens.model=~strata(zyg))
##' pconc <- predict(bcif1,newdata)
##'
##' ## marginal estimates
##' mbcif1 <- binreg(Event(time,status)~cluster(id), data=prt, time=80, cause=2)
##' mc <- predict(mbcif1,newdata)
##' mc
##'
##' cse <- binregCasewise(bcif1,mbcif1)
##' cse
##' @export
binregCasewise <- function(concbreg,margbreg,zygs=c("DZ","MZ"),newdata=NULL,...)
{# {{{
if (is.null(newdata)) newdata <- data.frame(zyg=zygs,id=1)
pconc <- predict(concbreg,newdata)
pmarg <- predict(margbreg,newdata)
f <- function(p) {concbreg$coef <- p; return(log(predict(concbreg,newdata)[,1]))}
fcase <- function(p) {
concbreg$coef <- p[1:2];
margbreg$coef <- p[3];
res <- (log(predict(concbreg,newdata)[,1]) - log(predict(margbreg,newdata)[,1]))
return(res)
}
margiid <- margbreg$iid
conciid <- matrix(0,nrow(margiid),nrow(pconc))
wiid <- which(concbreg$iid.origid %in% margbreg$iid.origid)
conciid[wiid,] <- concbreg$iid
iids <- cbind(conciid,margiid)
vcov <- crossprod(iids)
dd <- estimate(coef=c(concbreg$coef,margbreg$coef),vcov=vcov,f=fcase,...)
expcoef <- exp(dd$coefma[,c(1,3:4)])
res <- list(coef=expcoef,logcoef=dd)
return(res)
}# }}}