From: Chiaying Lin <*cylin292*>

Date: Tue, 8 Sep 2015 22:32:04 +0800

Dear NMusers,

I'm a NM beginner modeling for a monoclonal antibody. Below is 12

subjects's individual predicted plot , it seems that nonlinear

elimination (Michaelis-Menten) model can fit the data well. However, when

running the control stream (two compartment models with linear and

non-linear elimination), the resulting Q THETA SE% and OMEGA SE% are quite

large even thought successful minimization.From the Correlation Matrix , I

find that THETA (1) (VMAX) and THETA(2) (KM)are highly correlated (9.67E-01).

Another problem is the .fit file can't be produced.

I have tried various error models and initial estimations, however, none

had better improvement and often got error message (error 134

or parameter estimate is near its boundary). Does re-parameterization

helpful ? or need to change to other models?

Any suggestions for solving the problems are highly appreciated.

===================================================================

THETA: VMAX KM V1 V2 Q

ETA: BSVV1

ERR: ERR

run2.3.lst -167.989 eval=237 sig=+3.5 sub=12 obs=100 CCIL=YNYN NVI2.0 PV2.0

THETA = 7.35 59.6 3.68 0.87 0.0538

ETASD = 0.0676018

ERRSD = 0.25632

THETA:se% = 8.4 22.5 5.8 32.3 80.3

OMEGA:se% = 194.5

SIGMA:se% = 20.1

MINIMIZATION SUCCESSFUL

P VAL.: 0.10E+01

Ttot 0:44.28 Test 0:44.28 Tcov 0 Ttcl 0:0.44

$PROBLEM BASE MODEL

$DATA ..\PKDATA.CSV IGNORE=#

$INPUT ID TIME AMT NDV=DROP LNDV=DV EVID MDV CMT BW AGE DOSE RATE DUR

$SUBROUTINE ADVAN8 TOL=6

$MODEL

COMP=(CENTRAL, DEFOBS, DEFDOSE)

COMP=(PERIPH)

$PK

TVVMAX=THETA(1)

VMAX=TVVMAX

KM=THETA(2)

TVV1=THETA(3)

V1=TVV1*EXP(ETA(1))

TVV2=THETA(4)

V2=TVV2

Q=THETA(5)

S1=V1

D1=DUR

$DES

C1=A(1)/V1

C2=A(2)/V2

DADT(1)=-VMAX*C1/(KM+C1)-(Q*C1)+(Q*C2)

DADT(2)=(Q*C1)-(Q*C2)

$ERROR

IPRED=LOG(F)

IRES=DV-IPRED

IWRES=IRES/IPRED

Y=IPRED+ERR(1)

$THETA

(0.1, 7, ) ;VMAX

(1, 57,300) ;KM

(0, 4, 6) ;V1

(0, 1, 6) ;V2

(0.001, 0.003, ) ;Q

$OMEGA

0.001 ; BSVV1

$SIGMA

0.5 ;ERR

$ESTIMATION METHOD=1 INTERACTION MAXEVALS=9999 POSTHOC NOABORT NSIG=3

$COVARIANCE UNCONDITIONAL SLOW PRINT=E

$TABLE ID TIME DV IPRED VMAX KM V1 V2 Q ETA1 NOPRINT ONEHEADER FILE=

run2.3.fit

$TABLE ID TIME AMT IPRED IWRES NOPRINT ONEHEADER FILE=sdtab1

$TABLE ID VMAX KM V1 V2 Q ETA1 NOPRINT ONEHEADER FILE=patab1

$TABLE ID BW AGE NOPRINT ONEHEADER FILE=cotab1

Received on Tue Sep 08 2015 - 10:32:04 EDT

Date: Tue, 8 Sep 2015 22:32:04 +0800

Dear NMusers,

I'm a NM beginner modeling for a monoclonal antibody. Below is 12

subjects's individual predicted plot , it seems that nonlinear

elimination (Michaelis-Menten) model can fit the data well. However, when

running the control stream (two compartment models with linear and

non-linear elimination), the resulting Q THETA SE% and OMEGA SE% are quite

large even thought successful minimization.From the Correlation Matrix , I

find that THETA (1) (VMAX) and THETA(2) (KM)are highly correlated (9.67E-01).

Another problem is the .fit file can't be produced.

I have tried various error models and initial estimations, however, none

had better improvement and often got error message (error 134

or parameter estimate is near its boundary). Does re-parameterization

helpful ? or need to change to other models?

Any suggestions for solving the problems are highly appreciated.

===================================================================

THETA: VMAX KM V1 V2 Q

ETA: BSVV1

ERR: ERR

run2.3.lst -167.989 eval=237 sig=+3.5 sub=12 obs=100 CCIL=YNYN NVI2.0 PV2.0

THETA = 7.35 59.6 3.68 0.87 0.0538

ETASD = 0.0676018

ERRSD = 0.25632

THETA:se% = 8.4 22.5 5.8 32.3 80.3

OMEGA:se% = 194.5

SIGMA:se% = 20.1

MINIMIZATION SUCCESSFUL

P VAL.: 0.10E+01

Ttot 0:44.28 Test 0:44.28 Tcov 0 Ttcl 0:0.44

$PROBLEM BASE MODEL

$DATA ..\PKDATA.CSV IGNORE=#

$INPUT ID TIME AMT NDV=DROP LNDV=DV EVID MDV CMT BW AGE DOSE RATE DUR

$SUBROUTINE ADVAN8 TOL=6

$MODEL

COMP=(CENTRAL, DEFOBS, DEFDOSE)

COMP=(PERIPH)

$PK

TVVMAX=THETA(1)

VMAX=TVVMAX

KM=THETA(2)

TVV1=THETA(3)

V1=TVV1*EXP(ETA(1))

TVV2=THETA(4)

V2=TVV2

Q=THETA(5)

S1=V1

D1=DUR

$DES

C1=A(1)/V1

C2=A(2)/V2

DADT(1)=-VMAX*C1/(KM+C1)-(Q*C1)+(Q*C2)

DADT(2)=(Q*C1)-(Q*C2)

$ERROR

IPRED=LOG(F)

IRES=DV-IPRED

IWRES=IRES/IPRED

Y=IPRED+ERR(1)

$THETA

(0.1, 7, ) ;VMAX

(1, 57,300) ;KM

(0, 4, 6) ;V1

(0, 1, 6) ;V2

(0.001, 0.003, ) ;Q

$OMEGA

0.001 ; BSVV1

$SIGMA

0.5 ;ERR

$ESTIMATION METHOD=1 INTERACTION MAXEVALS=9999 POSTHOC NOABORT NSIG=3

$COVARIANCE UNCONDITIONAL SLOW PRINT=E

$TABLE ID TIME DV IPRED VMAX KM V1 V2 Q ETA1 NOPRINT ONEHEADER FILE=

run2.3.fit

$TABLE ID TIME AMT IPRED IWRES NOPRINT ONEHEADER FILE=sdtab1

$TABLE ID VMAX KM V1 V2 Q ETA1 NOPRINT ONEHEADER FILE=patab1

$TABLE ID BW AGE NOPRINT ONEHEADER FILE=cotab1

Received on Tue Sep 08 2015 - 10:32:04 EDT