[R] Actual vs. predicted survival times
alanm (Alan Mitchell)
alanm at crab.org
Wed Feb 8 15:43:08 CET 2012
Have you tried plotting obvserved survival against X*Beta? I believe the usual predictions from a cox model are just monotonic transformations of this.
-Alan
-----Original Message-----
From: Bonnett, Laura [mailto:L.J.Bonnett at liverpool.ac.uk]
Sent: Wed 2/8/2012 1:52 AM
To: 'David Winsemius'
Cc: 'r-help at r-project.org'
Subject: Re: [R] Actual vs. predicted survival times
Dear David,
Thank you for your input. I was under the impression that it was not possible to obtain survival times using the predict.coxph function. In a post to the R help group I noticed someone had suggested calculating survival times in the manner I have used although obviously this is not appropriate.
Are you able to provide any suggestions as to how I might achieve the graph of actual vs. Predicted survival times?
Many thanks,
Laura
-----Original Message-----
From: David Winsemius [mailto:dwinsemius at comcast.net]
Sent: 07 February 2012 16:59
To: Bonnett, Laura
Cc: 'r-help at r-project.org'
Subject: Re: [R] Actual vs. predicted survival times
On Feb 7, 2012, at 10:32 AM, Bonnett, Laura wrote:
> Dear R-help,
>
> I am using R 2.14.1 on Windows 7.
>
> I would like to produce a plot like the attached - although
> simplified to actual vs. Predicted survival time with distinguishing
> marks for censored and observed points. I have a dataset and have
> fitted a Cox model to it. In an attempt to visualise how accurate
> the model is it would be ideal if I could plot the actual survival
> times against the predicted survival times.
>
> I have been looking on the internet to see if there are ways to do
> this in R. The only post I found (https://stat.ethz.ch/pipermail/r-help/2009-February/189888.html
> ) that seemed directly relevant suggested that I shouldn't be
> generating survival times at all. Given that, I was concerned about
> proceeding but I would like to have access to a plot to make a
> decision on its usefulness.
>
> I appreciate that there are predict.coxph and predict.cph options
> available to me.
>
> My first attempt was as follows:
>
> # fit Cox model #
> coxfita = coxph(Surv(tsecond,seccens)~stroke(smess)+rels(smess)
> +asleep(smess)+eeg1(smess)+eeg2(smess)+ct1(smess)+ct2(smess)
> +treat(smess),data=smess)
>
> # Find censored and observed groups #
> messcens <- subset(smess,seccens==1)
> messobs <- subset(smess,seccens==0)
>
> # Obtain predicted survival times #
> explp <- exp(predict(coxfita,type="lp"))
That gives you relative risks, not survival times.
> explp2 <- mean(ssmess$tsecond,na.rm=TRUE)*explp
Why are you multiplying times by relative risks? That makes no sense.
--
David.
> smess2 <- data.frame(ssmess,explp2)
>
> # Find censored and observed groups #
> smesscens <- subset(smess2,seccens==1)
> smessobs <- subset(smess2,seccens==0)
>
> # Produce plot #
> plot(smesscens$explp2,messcens$tsecond,pch=4,col="blue",ylab="Actual
> Survival Time",xlab="Predicted Survival Time",main="Survival
> Times",xlim=c(0,3500),ylim=c(0,3500))
> points(smessobs$explp2,messobs$tsecond,pch=4,col="red")
>
> This leads to the attached plot. It doesn't seem correct though as
> the predicted times all start over 500 days.
>
> Any suggestions would be very welcome.
>
> Many thanks,
> Laura
>
> <Actual vs. Survival
> LJB.pdf>______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
David Winsemius, MD
West Hartford, CT
More information about the R-help
mailing list