$SIZES LIM6=1500 
$PROBLEM MIT COVARIATE MODEL, TREATMENT AFFECTS MARGINATION AND DEMARGINATION.

$INPUT STUDY ID TIME CP=DV DOSE=AMT RATE DURATION CMT EVID MDV CENS YTYPE TYPE DBY DRUG DCH DTYP 
       BIO V2 ADA KE VM KM KA K23 K32 MTT ANC WT DVVSE ORDER PCTCH RATIO PCMT TREATMENTN STRATA DVL=DROP LIMIT=DROP

  ;EVID: 0-OBSERVATION, 1-DOSE, 2-DUMMY, 4-DOSE & RESET.
$DATA "../DA/D8806002.DAS" WIDE IGNORE=# 
$SUBROUTINES  ADVAN14  TOL=9    
  ;TOL - THE NUMBER OF ACCURATE DIGITS
$MODEL COMP=(INJ) 
       COMP=(BLOOD) 
       COMP=(AUC)
       COMP=(KLETKI)
       COMP=(RT1) COMP=(RT2) COMP=(RT3) 
       COMP=(PD1) COMP=(PD2) COMP=(PD3) 

$THI
  THETA=LOG(THETAI(1:7))
$THR
  THETAR=EXP(THETA(1:7)) 

$PK
  MU_1  =THETA(1)
  MU_2  =THETA(2)
  MU_3  =THETA(3)
  MU_4  =THETA(4)
  MU_5  =THETA(5)

  MU_6  =THETA(6)
  MU_7  =THETA(7)
  MU_8  =THETA(8)
  MU_9  =THETA(9)
  MU_10 =THETA(10)

  BTO   = DEXP(MU_7+ETA(7))

  INDI = DRUG-1

  KTR  = 4/MTT

  F1   = BIO
 
  KOUT = 0.877
 
  S2=V2    ;S2 IS THE AMOUNT A IN THE OBSERVATION COMPARTMENT
           ;AT THE TIME OF OBSERVATION, DIVIDED BY THE 
           ;VALUE OF A PARAMETER S, IS USED AS THE PREDICTION. 


  BETA_ANConA80 = DEXP((MU_8+ETA(8))*DLOG(ANC/4.75))
  A80           = DEXP(MU_1+ETA(1))*BETA_ANConA80  

  BETA_ANConB = DEXP((MU_10+ETA(10))*DLOG(ANC/4.75))
  B    = DEXP(MU_2+ETA(2))*BETA_ANConB

  GA   = DEXP(MU_3+ETA(3))
  D890 = DEXP(MU_4+ETA(4))
  KTRN = DEXP(MU_5+ETA(5))

  D980 = D890

  BETA_DRGonEC50 = DEXP((MU_9+ETA(9))*INDI)
  EC50           = DEXP(MU_6+ETA(6))*BETA_DRGonEC50

  KIN = A80*KOUT*(D890+D980+KOUT)/(D980+KOUT)

  A_0(1)  = 0
  A_0(2)  = 0
  A_0(3)  = 0
  A_0(4)  = 0
  A_0(5)  = 0
  A_0(6)  = 0
  A_0(7)  = 0
  A_0(8)  = A80
  A_0(9)  = A80*D890/(D980+KOUT)
  A_0(10) = KIN/KTRN


$ERROR
  CALLFL=0
  IPRE    = F
  SIG     = THETA(11)
  LNDV    = DLOG(DV)
  LNIPRE  = DLOG(IPRE)
  DV01    =(LNDV-LNIPRE)/SIG
  LNSIG   = DLOG(SIG)
 
  LL2 = DV01*DV01 + 2*(LNDV + LNSIG) + 1.837876
  Y  = LL2

  IPRED=IPRE 
  IWRES=DV01 

$DES
  IND=DTYP
  DADT(1) =-KTR*A(1)*IND                                                             ;INJ
  DADT(2) =-A(2)*KE-K23*A(2)+K32*A(4)+KA*A(7)*IND-A(2)*VM/(KM+A(2)/V2)               ;KROV
  DADT(3) = A(2)/V2                                                                  ;AUC
  DADT(4) = K23*A(2)-K32*A(4)   

  DADT(5) =-KTR*A(5) +KTR*A(1)
  DADT(6) =-KTR*A(6) +KTR*A(5)
  DADT(7) =-KA *A(7) +KTR*A(6)

  C2      = A(2)/V2

  mult = (1 + B*C2/(EC50+C2))
  D89 = D890*mult
  D98 = D980/mult

  multTO = (1 + BTO*C2/(EC50+C2))

  DADT(8) =-D89*A(8) + D98*A(9)     + KTRN*A(10) - KOUT*A(8)*multTO 
  DADT(9) = D89*A(8) - D98*A(9)                  - KOUT*A(9)*multTO 

  DADT(10)=-KTRN*A(10)   + KIN*multTO    


  A1 =A(1)
  A2 =A(2)/V2
  A3 =A(3)
  A4 =A(4)
  A5 =A(5)
  A6 =A(6)
  A7 =A(7)
  A8 =A(8)
  A9 =A(9)
  A10=A(10)

  CHANGE = A8-A80
  PCTCHANGE = CHANGE/A80*100


$THETA (1E-5,4.02,100)      ; POP_A80
$THETA (0.03,1.22,100)      ; POP_B 
$THETA (1 FIXED)            ; POP_GA 
$THETA (1E-5,1.80,100)      ; POP_D890 
$THETA (1E-4,0.248,100)     ; POP_KTRN

$THETA (0.01,0.216,100)     ; POP_EC50 
$THETA (0.1,1.375,100)      ; POP_BTO  
$THETA (-100,0.81,100)      ; POP_BETA_ANConA80 
$THETA (-100,1.56,100)      ; POP_BETA_DRGonEC50 
$THETA (-100,0.5,100)       ; POP_BETA_ANConB 

$THETA (0.00001,0.244,10)   ; POP_SD1




$OMEGA DIAGONAL(2)  0.0133                        ;ETA_A80
                    0.919                         ;ETA_B

$OMEGA DIAGONAL(8) 0 FIX                              ; ETA_GA
                   0 FIX                              ; ETA_D890
                   0 FIX                              ; ETA_KTRN
                   0 FIX                              ; ETA_EC50
                   0 FIX                              ; ETA_BTO
                   0 FIX                              ; ETA_BETA_ANConA80
                   0 FIX                              ; ETA_BETA_DRGonEC50
                   0 FIX                              ; ETA_BETA_ANConB


$EST NOABORT PRINT=1 NBURN=3000 NITER=500 CONSTRAIN=7
  METHOD=SAEM  POSTHOC ;INTERACTION  
  CTYPE=3 CINTERVAL=10 
  LAPLACIAN GRD=TG(1-10):TS(11) DERCONT=1 -2LL
  SIGL=8 SIGLO=8

$EST METHOD=IMP EONLY=1 ISAMPLE=1000 NITER=20 MAPITER=0

$COV MATRIX=R PRINT=E UNCONDITIONAL ;SLOW