Re: Output predicted slopes for each individual in proc mixed

Abishekaa

is it possible to output predicted individual-level slopes from proc mixed? Here is my current code. Thanks!

proc mixed data=all covtest cl;
  class ptID;
  model value=visitYr / s residual outp=pred;
  random int visitYr / type=un subject=ptID g gcorr v;
  ods output CovParms=covout solutionr=slopes;;
 run;

ballardw

See if this post is helpful: https://communities.sas.com/t5/Statistical-Procedures/predicted-value-proc-mixed/td-p/56468

data_null__

Show us some of this solutionr=slopes

@Abishekaa wrote:

is it possible to output predicted individual-level slopes from proc mixed? Here is my current code. Thanks!
proc mixed data=all covtest cl;
  class ptID;
  model value=visitYr / s residual outp=pred;
  random int visitYr / type=un subject=ptID g gcorr v;
  ods output CovParms=covout solutionr=slopes;;
 run;

Ksharp

You could use SOLUTION option of RANDOM to get them.

proc mixed data=sashelp.heart(obs=1000) covtest cl;
  class bp_status;
  model weight=height / s residual outp=pred;
  random int height / type=un subject=bp_status g gcorr v  solution;
  ods output CovParms=covout solutionr=slopes;;
 run;

StatsMan

The SOLUTION option on the RANDOM statement gives you the random deviation of each individual's slope from the overall slope. The overall slope comes from the coefficient for HEIGHT given in the S option for the MODEL statement. To get individual slopes, combine the results of the S option on the RANDOM statement and the S option on the MODEL statement.

jiltao

The following usage note provides sample programs for obtaining subject-specific intercept and slope --

http://support.sas.com/kb/37109

The following note shows how to plot these subject-specific models --

http://support.sas.com/kb/37110

Hope this helps,

Jill