257
fitting of the same data using the "Multi" program (Appendix I) of
54 .
Yamaoka et al. By giving weights of 0,1, and 2 for the experimental
Yi values, the estimated parameters of the plasma concentration time
profile of 1 in dog C at 1.439 mg/kg dose and their respective minimum
sum of squares are given in Table Al (Fig. Al, also Fig. 8 in main text
for weight=l). Notice that when the sum of squares is a minimum with
equal (or 0) weighting, the fit completely missed all the terminal
points (Fig. Al-d).
Figure Al represents a type of fit where plasma concentrations were
transformed to their respective natural logarithms and then fitted by
"Multi" (Appendix I) with equal (or 0) weighting.
Except for equal (or 0) weighting, all other methods of weighting
gave parameters that were reasonably similar (Table Al). The unweighted
sum of squares calculated for various weighting was a minimum (excluding
0 weight) for weight=l (See Fig. 8 in main text and also Table Al). Thus
inverse plasma concentrations were chosen as the appropriate weighting
factor in all fittings of the plasma concentation-time profile of 1 in
dogs.