___________________________________________________________________
 |                                                                 |
 |                             BIND EXAMPLE                        |
 |_________________________________________________________________|

 EXAMPLES OF THE USE OF $BIND

 Case 1.

 Suppose a fragment of the dose records looks like this:

        TIME AMT PREP    X
          0  100    4    .
         10  100    1   90
         20    0    0   80

 Suppose $INPUT and $PK include this code:

 $INPUT ID TIME AMT DV PREP X
 $PK
     F1=1
     IF (PREP.EQ.4) F1=THETA(3)
     ALAG1=THETA(5)
     CL=THETA(1)*X*EXP(ETA(1))

 It is intended that theta(3) be  the  ratio  of  bioavailabilities  of
 preparation  4 to preparation 1. However, calls to PK are made only at
 event times 0, 10 and 20. Suppose the value of theta(5) (lag time)  is
 5.   After  5 time units, the first dose enters the system.  The value
 of PREP (dose preparation type) is taken from the event record at time
 10, which is 1. F1 is computed incorrectly.  If the estimation step is
 run, the gradient with theta(3) will be 0, indicating the problem.

 Case 2.

 If lag time is not modeled (no ALAG1),  but  the  doses  are  multiple
 (ADDL>0,  II>0), a more subtle error can occur.  Suppose a fragment of
 the data records looks like this:

        TIME AMT PREP    X  II  ADDL
          0  100    4    .   5     1
         10  100    1   90   5     1
         20    0    0   80   .     .

 Suppose $INPUT and $PK include this code:

 $INPUT ID TIME AMT DV PREP X II ADDL
 $PK
     F1=1
     IF (PREP.EQ.4) F1=THETA(3)
     CL=THETA(1)*X*EXP(ETA(1))

 F1 is computed correctly for the initiating dose records  at  times  0
 and  10,  and  for the additional dose at time 15, but incorrectly for
 the additional dose at time 5.  Theta(3) applies  only  to  the  first
 dose  having  PRED=4.  (The gradient for theta(3) is not 0.)  The same
 problem could have occurred with the dose having PREP=1,  except  that
 by  an  accident  of  coding, PREP=0 (on the record having TIME=20) is
 treated the same as PREP=1, so that the additional dose at time 15 has
 F1=1 as it should.

 On the other hand, the values of X are presumably recorded so that  90
 is  valid  for  the  time 0 through 10 and 80 is valid for the time 10
 through 20. Therefore, the model for CL is correct in both examples.

 In either of these two examples, the model can  be  corrected  by  the
 insertion of $BIND and CALLFL=-2:

 $INPUT ID TIME AMT DV PREP X II ADDL ; II & ADDL only in case 2
 $BIND   -    -   -  - DOSE NEXT
 $PK
     CALLFL=-2

 Because of CALLFL=-2 in the abbreviated code, additional calls  to  PK
 are made at the nonevent dose times (5 in case 1; 5 and 15 in case 2).
 Because of the $BIND record, PREP has the value at  these  calls  from
 the  initiating  dose  record (4 at time 5, 1 at time 15).  F1 is com-
 puted appropriately for the dose preparation. X still has values  from
 the next event record, so the computation of CL is unchanged, e.g., it
 is the same from time=0 to time=5 as it is from  time=5 to time=10.

 (See $BIND).

REFERENCES: Guide IV Section V.C.2, V.C.5

REFERENCES: Guide VI Section III.B.2, V.K


  
Go to main index.
  

    Created by nmhelp2html v. 1.0 written by Niclas Jonsson (Modified by AJB 5/2006,11/2007)