RE: Model building advice

From: "Eleveld, DJ" d.j.eleveld@anest.umcg.nl Subject: RE: [NMusers] Model building advice Date: Thu, April 28, 2005 7:04 am Thank you for your suggestions... Serge, i'm not exactly sure what you mean by 'node' in the intergration routine, do you mean event? I tried to use the bioavalibility parameter as Leonid Gibiansky suggested in a mail to me. I added dose events (fixed size) at the appropriate times into the appropriate compartment and manipulate the bioavalibility to change the size of step but this does not work. I am not allowed to change bioavalability constants in the $DES code. Neither may I change bioavailability in the $ERROR code. When using fixed dose sizes and and fixed bioavalibility then all the step changes will be the same. But I need the size of the step change to change during the integration. To better describe what I mean by step change in state variable, essentially what I would like to have in my $DES and $ERROR code is: $DES ... DADT(5)=-POTK*A(5) ; Decay bewteen twitches ... $ERROR CALLFL=0 ... STEP=.... A(5)=A(5)+STEP ; Produce the step change But I cannot manipulate A(5) within the $ERROR code. The relevant part of the control stream is: (error line at the bottom) -------Control stream-------- $PROB Potentiation fitting $DATA potent.da2 $INPUT ID TIME DV CFLG MDV AMT RATE CMT $SUBROUTINES ADVAN9 TOL=3 $MODEL NCOMPARTMENTS=5 NPARAMETERS=12 COMP(CENTRAL DEFDOSE) COMP(PERIF1 NOOFF NODOSE) COMP(PERIF2 NOOFF NODOSE) COMP(EFFECT NOOFF NODOSE) COMP(POTENT NOOFF) $PK CALLFL=1 V1=THETA(1)*EXP(ETA(1)) V2=THETA(2)*EXP(ETA(2)) V3=THETA(3)*EXP(ETA(3)) CL=THETA(4)*EXP(ETA(4)) Q3=THETA(6)*EXP(ETA(6)) Q2=(THETA(2)*(THETA(6)/THETA(3) + THETA(5)))*EXP(ETA(5)) S1=V1 KEO=THETA(7)*EXP(ETA(7)) EC50=THETA(8)*EXP(ETA(8)) GAMM=THETA(9)*EXP(ETA(9)) POTR=THETA(10)+ETA(10) ; r has normal dist POTK=THETA(11)*EXP(ETA(11)) SCAL=THETA(12)*EXP(ETA(12)) $DES C1=A(1)/V1 ; Conc in V1 C2=A(2)/V2 ; Conc in V2 C3=A(3)/V3 ; Conc in V3 DADT(1)=Q2*(C2-C1)+Q3*(C3-C1)-CL*C1 ; Change in V1 DADT(2)=Q2*(C1-C2) ; Change in V2 DADT(3)=Q3*(C1-C3) ; Change in V3 DADT(4)=(C1-A(4))*KEO ; Effect compartment conc DADT(5)=-POTK*A(5) ; Decay potentiation $ERROR CALLFL=0 PTMP=A(5) CPRE=A(1)/V1*EXP(ERR(1)) PD1=A(4)**GAMM ; Sigmoidal Emax model NMB=PD1/(PD1+(EC50**GAMM)) RPRE=SCAL*(1+A(5))*(1-NMB)+ERR(2) ; Twitch in % IF (DV.EQ.-999) RPRE=-999+ERR(2) Y=CFLG*CPRE+(1-CFLG)*RPRE ; *** Step change once per twitch *** ERROR *** IF (CFLG.NE.0) A(5)=A(5)+POTR*(1-NMB) I hope this makes more clear what I mean by 'step' change in state variable. Does anyone have any advice as to how I can achieve this kind of effect? Thanks very much, Doug Eleveld