Re: [NMusers] help on diagnostic of a problem -- about M3 method for LLOQ:

ominator to be variance.    after changing it, the model works fine n=
ow.    I meant to have different additive error for study greater=
 than 3. When I used log transformed data, I didn't see the need to have co=
mbination error model, but, I still kept estimating different err for studi=
es. ________________________________ From: "Ahn, Jae Eun" <jaeeun=
.ahn_at_pfizer.com> To: Ethan Wu <ethan.wu75_at_yahoo.com>; nmusers_at_globomaxnm.=
com Sent: Thu, February 11, 2010 12:06:49 PM Subject: RE: [NMusers] hel=
p on diagnostic of a problem -- about M3 method for LLOQ: Dear Etha=
n,   The denominator of DUM should be in the standard deviation level=
, not variance, so it should be W     = SQRT(THETA(7)**2*FLAG+T=
HETA(8)**2*(1-FLAG)+THETA(9)**2*IPRED**2) IF (TYPE.EQ.2.AND.CMT.EQ.2) DUM=
=(LOQ-IPRED)/W   Note that the way you coded DUM is equivalent to D=
UM=(LOQ-IPRED)/W**2   In your code for log-transformed data, I see =
that the denominator is specified as a standard deviation.   However,=
 it is not clear to me whether you want to use a different magnitude of pro=
portional error or additive error for studies whose number is greater than =
3. In the code for natural DV, THETA(8) is for the different additive error=
 of such studies, whereas THETA(8) in the log DV corresponds to the differe=
nt proportional error.   Best regards, Jae Eun   _________=
_______________________ From:owner-nmusers_at_globomaxnm.com [mailto:owne=
r-nmusers_at_globomaxnm.com] On Behalf Of Ethan Wu Sent: Thursday, February =
11, 2010 10:53 AM To: nmusers_at_globomaxnm.com Subject: [NMusers] help on=
 diagnostic of a problem -- about M3 method for LLOQ:   Dear users,=

his error message: 0PROGRAM TERMINATED BY OBJ ERROR IN CLIK WITH INDIVI=
DUAL 2 ID=0.19000000E+02 CONDITIONAL LIKELIHOOD FOR SOME OBSERVATION ES=
TIMATED TO BE NEGLIGIBLE   note that ID 19 is the first subject in th=
e dataset. this is how I specify the $ERR (a combinative erro) and $EST:=

S  = DV-IPRED    FLAG=1    IF(STDY.GT.3) FLAG=0  =
  DEL =0    W     = SQRT(THETA(7)**2*FLAG+THETA(8)**2*(=
1-FLAG)+THETA(9)**2*IPRED**2)    IF(W.EQ.0) DEL =1    IWRES =
= IRES/( W+DEL ) IF (TYPE.EQ.2.AND.CMT.EQ.2) DUM=(LOQ-IPRED)/(THETA(7=
)**2*FLAG+THETA(8)**2*(1-FLAG)+THETA(9)**2*IPRED**2)  CUMD=PHI(DUM)=

) THEN F_FLAG=0  Y = IPRED+W*EPS(1)  ENDIF $ESTIMATION MET=
HOD=1 LAPLACIAN  MAXEVAL=9999 PRINT=10 NOABORT or this: $ESTIM=
ATION METHOD=1 INT LAPLACIAN  MAXEVAL=9999 PRINT=10 NOABORT __=
___ Interstingly, when I use Log-transformed data, the model would run fi=
ne. this is how I specify $ERR (an additive error) and $EST, all the othe=
r aspects of model are the same between two tries. Could someone help me =
understand what is happening? Thanks. _____ $ERROR    DEL    =
= 0    IF(F.LE.0) DEL = 1    IPRED = LOG(F+ DEL )  =
  IRES  = DV-IPRED    FLAG=1    IF(STDY.GT.3) FLAG=0=

   DEL1=0    IF(W.EQ.0)DEL1=1    IWRES = IRES/( W+DEL=
1 ) IF (TYPE.EQ.2.AND.CMT.EQ.2) DUM  = (LOG(LOQ)-IPRED)/(THETA(7)*FLA=
G+THETA(8)*(1-FLAG))  CUMD=PHI(DUM) IF (TYPE.EQ.2) THEN  F_FLA=
G=1 Y=CUMD ENDIF IF (TYPE.EQ.1) THEN F_FLAG=0 Y=IPRED +=
EPS(1)* W ENDIF   $ESTIMATION METHOD=1 LAPLACIAN MAXEVAL=9999 P=
RINT=10 NOABORT
Received on Thu Feb 11 2010 - 13:40:40 EST