Predictive survival probability IN PHREG

Reply
Super Contributor
Posts: 297

Predictive survival probability IN PHREG

Hi How do I get the predictive survival probability from this?;

proc phreg data=ph;

model zz*status(0)=x1/RL TIES=EFRON;

run;

Super User
Posts: 17,829

Re: Predictive survival probability IN PHREG

If you want it for specific variables see the following in the documentation.

Example 64.8 Survivor Function Estimates for Specific Covariate Values

If at specific time points, but general see the output and the timelist option.

Super Contributor
Posts: 297

Re: Predictive survival probability IN PHREG

Thanks, however I cant find the predictive value. There is no data set Pred1 in my work directory

 

data Inrisks;

  length Id $20;

  input LogBUN HGB Id $12-31;

  datalines;

  1.00 10.0 logBUN=1.0 HGB=10

  1.80 12.0 logBUN=1.8 HGB=12

ods graphics on;

  proc phreg data=Myeloma plots(overlay)=survival;

  model Time*VStatus(0)=LogBUN HGB;

  baseline covariates=Inrisks out=Pred1 survival=_all_ / rowid=Id;

  run;

  ods graphics off;

Super User
Posts: 17,829

Re: Predictive survival probability IN PHREG

What does your log say?

Super Contributor
Posts: 297

Re: Predictive survival probability IN PHREG

No error message, when I used covariates=ph, it multiple data point in SURVs 100 times(the sample size)

6924  proc phreg data=ph;

6925  model zz*status(0)=x1/ TIES=EFRON;

6926  baseline out=survs   out=Pred1 survival=s;

WARNING: Ignoring second data set reference.

6927  run;

Output Added:

-------------

Name:       ModelInfo

Label:      Model Information

Template:   Stat.Phreg.ModelInfo

Path:       Phreg.ModelInfo

-------------

Output Added:

-------------

Name:       NObs

Label:      Number of Observations

Template:   Stat.Phreg.NObs

Path:       Phreg.NObs

-------------

Output Added:

-------------

Name:       CensoredSummary

Label:      Summary of Event and Censored Observations

Template:   Stat.Phreg.CensoredSummary

Path:       Phreg.CensoredSummary

-------------

Output Added:

-------------

Name:       ConvergenceStatus

Label:      Convergence Status

Template:   Stat.Phreg.ConvergenceStatus

Path:       Phreg.ConvergenceStatus

-------------

NOTE: Convergence criterion (GCONV=1E-8) satisfied.

Output Added:

-------------

Name:       FitStatistics

Label:      Model Fit Statistics

Template:   Stat.Phreg.FitStatistics

Path:       Phreg.FitStatistics

-------------

Output Added:

-------------

Name:       GlobalTests

Label:      Test of Global Null Hypothesis

Template:   Stat.Phreg.GlobalTests

Path:       Phreg.GlobalTests

-------------

Output Added:

-------------

Name:       ParameterEstimates

Label:      Maximum Likelihood Estimates of Model Parameters

Template:   Stat.Phreg.ParameterEstimates

Path:       Phreg.ParameterEstimates

-------------

NOTE: The data set WORK.SURVS has 84 observations and 3 variables.

NOTE: PROCEDURE PHREG used (Total process time):

      real time           0.30 seconds

      cpu time            0.15 seconds

Super User
Posts: 17,829

Re: Predictive survival probability IN PHREG

desireatem wrote:

No error message, when I used covariates=ph, it multiple data point in SURVs 100 times(the sample size)

6924  proc phreg data=ph;

6925  model zz*status(0)=x1/ TIES=EFRON;

6926  baseline out=survs   out=Pred1 survival=s;

WARNING: Ignoring second data set reference.

6927  run;

...

NOTE: The data set WORK.SURVS has 84 observations and 3 variables.

NOTE: PROCEDURE PHREG used (Total process time):

      real time           0.30 seconds

      cpu time            0.15 seconds

Notice the warning and the note at the end. Check the survs dataset and see what it has.

Note that you didn't specify the covariates as in the example.

Super Contributor
Posts: 297

Re: Predictive survival probability IN PHREG

The survs data is fine, got all what it suppose to have. There is no pred1 data

Super User
Posts: 17,829

Re: Predictive survival probability IN PHREG

No there isn't. You specified out twice, once with survs and once with pred1 and SAS put a warning, it would only do use the first one.

WARNING: Ignoring second data set reference.

Follow the example, set out. Specify your covariates and then you'll get the estimates at covariate levels, otherwise the survs dataset will have the estimate at different time intervals.  If you want a specific time interval specify the timelist. 

Ask a Question
Discussion stats
  • 7 replies
  • 486 views
  • 0 likes
  • 2 in conversation