Re: Standard error for estimates

From: Kazimierz H. Kozlowski Date: June 20, 2007 technical Source: mail-archive.com
Dear Navin, I suggest following codes, but you need carefuilly check it - I.m just on holidays and writing from my head only. $PROBLEM BIOAVAILABILITY MODEL $INPUT ID TIME AMT MDV DV EVID CMT $DATA file1.csv IGNORE=C $SUBROUTINE ADVAN8 TRANS1 TOL=7 $MODEL COMP=(DEPOT,DEFDOSE) ; ABSORPTION COMPARTMENT COMP=(CENTRAL, DEFOBS) ; PLASMA COMPARTMENT $PK F1=THETA(1)*ETA(1)+ETA(2) ; HETEROSCEDASTIC LINEAR MODEL OF INTERSUBJECTS VARIANCE - IF NOT GOOD PLEASE CORRECT KA=4.0 ; IDENTICAL KA PARAMETER FOR ALL PATIENTS IN [1/H] K=0.20 ; IDENTICAL K - ELIMINATION RATE CONSTANT IN [1/H] = BETA TVCL=THETA(2) ; TYPICAL VALUE CL = 75 ML/MIN/KG (FROM INTRAVENOUS STUDY ?) CL=TVCL*EXP(ETA(3)) ; INDYVIDUAL CL ACCORDING TO ETA (3) MODEL TVVC=TVCL/K ; TYPICAL VALUE VOLUME OF DISTRIBUTION IN [ML/KG] = 75 [ML/KG/H]/0.2 [1/H] = 375 [ML/KG] VC=TVVC*EXP(ETA(4)) ; INDYVIDUAL VC ACCORDING TO ETA (4) MODEL S2=VC/1000 ; S2 IS SCALED ACCORDING TO DOSE [MICROGRAMS/KG] AND OUTPUT IN [NG/ML] $ERROR IPRED=F IRES=DV-IPRED DEL=0 IF (IPRED.EQ.0) DEL=1 IWRES=(1-DEL)*IRES/(IPRED+DEL) Y=F+F*ERR(1) $DES DADT(1)=-KA*A(1) DADT(2)=KA*A(1)-CL/K*A(2) $THETA (0.01,0.5,0.8) ; TVF1 - THETA(1) (75 FIXED) ; TVCL -THETA(1) $OMEGA 0.02 0.1 0.1 0.05 ; ETA(1,2,3,4) - INTERSUBJECTS VARIANCES $SIGMA 0.07 ; ERR(1) - INTRASUBJECT VARIANCE $ESTIMATION METHOD=1 INTERACTION LAPLACE SIGDIGITS=6 MAXEVAL=9999 PRINT=5 MSF=SDMSFSAB.335 $COVARIANCE UNCONDITIONAL MATRIX=S $TABLE ID TIME AMT MDV DV EVID CMT IPRED IWRES IRES NOPRINT ONEHEADER FILE=SDTABSAB.335 sincerely Kazimierz H. Kozlowski navin goyal wrote: > Dear Kazimierz, > > I guess I did not quite follow your suggestion Sir. Could you please elaborate briefly on it. I think what I intended o do was havea fixed theta value and apply some variability ETA(3) and ETA(4) to those parameter estimates. I apologise for not following you correctly. I failed to understand the last line modification that you suggested. > > S2=VC; OUTPUT IN ng/ml =====> scale S2 vs DOSE > > Thanks a lot for your help >
Quoted reply history
> On 6/15/07, Kazimierz H. Kozlowski <[EMAIL PROTECTED] < mailto:[EMAIL PROTECTED]>> wrote: > > Dear Navin, > > You probably have double definition code for VC and also problem > that CL and VC are not constants between patients because of > ETA(3) and ETA(4) > > $PK > > F1=THETA(1)*EXP(ETA(1)+ETA(2)) > K12=4;Abso constant > TVCL=THETA(2);Clearance > CL=TVCL*EXP(ETA(3)) > TVVC=THETA(3);vol of dist of drug =====> delete > VC=TVVC*EXP(ETA(4)) =====> delete > K20=0.20;elim rate const of drug > VC=CL/K20 ====> OR rename VC1=CL/K20 > S2=VC; OUTPUT IN ng/ml =====> scale S2 vs DOSE > > sincerely > Kazimierz H. Kozlowski > > navin goyal wrote: > > > Hello everybody > > I am using Nonmem VI to model some Population Pk (10 subjects) > > data that I have. > > > > When I run this file.. I get the parameter estimates. > > However I fail to get any STANDARD ERROR of the estimates. > > I do sometimes get the R,S, T matrices when I change my initial > > estimates. > > Also the output file has following messages > > > > MINIMIZATION SUCCESSFUL > > MATRIX ALGORITHMICALLY SINGULAR > > AND ALGORITHMICALLY NON-POSITIVE-SEMIDEFINITE > > 0R MATRIX IS OUTPUT > > 0COVARIANCE STEP ABORTED > > > > My understanding (maybe I got it wrong) is that these R,S, T > > matrices are printed out when there are more than one solutions > > to the parameter estimates. > > is this right or does my control stream have some error within it. > > > > I checked the individual plots with the output data file. The > > individual and population predicted fits are quite decent.. > > > > It also gives me the following error message in the dos window > > > > (WARNING 50) OPTION "UNCONDITIONAL" OF $COVARIANCE RECORD IS > > IGNORED WITH > > THIS RELEASE OF NONMEM. > > > > Could someone kindly help me with what is the error in my control > > stream or how would I be able to get the Std error for the estimates > > > > $PROBLEM PK > > $INPUT ID TIME AMT MDV DV EVID CMT > > $DATA file1.csv IGNORE=C > > $SUBROUTINE ADVAN6 TRANS1 TOL=5 > > $MODEL > > COMP=(DEPOT,DEFDOSE); > > COMP=(CENTRAL);PLASMA > > > > $PK > > > > F1=THETA(1)*EXP(ETA(1)+ETA(2)) > > K12=4;Abso constant > > TVCL=THETA(2);Clearance > > CL=TVCL*EXP(ETA(3)) > > TVVC=THETA(3);vol of dist of drug > > VC=TVVC*EXP(ETA(4)) > > K20=0.20;elim rate const of drug > > VC=CL/K20 > > S2=VC; OUTPUT IN ng/ml > > > > $ERROR > > IPRED=F > > IRES=DV-IPRED > > DEL=0 > > IF (IPRED.EQ.0) DEL=1 > > IWRES=(1-DEL)*IRES/(IPRED+DEL) > > Y=F+F*ERR(1) > > > > $DES > > DADT(1)=-K12*A(1) > > DADT(2)=K12*A(1)-K20*A(2) > > > > $THETA > > (0.01,0.5,0.8); bioavailability > > (75 FIXED);CL > > (375 FIXED);VC > > > > $OMEGA > > > > $SIGMA > > > > $ESTIMATION METHOD=1 SIGDIGITS=2 MAXEVAL=9999 PRINT=5 POSTHOC > > $COVARIANCE UNCONDITIONAL > > > > $TABLE ID TIME AMT MDV DV EVID CMT IPRED IWRES IRES > > NOPRINT ONEHEADER FILE=sdtabsab335 > > > > thanks a lot
Jun 15, 2007 Navin Goyal Standard error for estimates
Jun 15, 2007 Kazimierz H. Kozlowski Re: Standard error for estimates
Jun 20, 2007 Kazimierz H. Kozlowski Re: Standard error for estimates