Remember to limit the time spent on this section.

Q 3.1 Briefly describe the parameter elimination rate constant. Give a definition, a working (useful) equation and units.

Elimination rate constant, kel. Units time^-1, e.g. hr^-1. First order rate constant. Relating rate of change with amount remaining, e.g. Cp. Independent of amount remaining, e.g. Cp. Working equations:

kel = (ln(Cp1)-ln(Cp2))/(t2-t1)

delta(Cp)/delta(t) = -kel x Cp

Note: Not asking about all parameters, just kel.

Q 3.2 Comment on the following results:

** FINAL PARAMETER VALUES ***

  #  Name                  Value       S.D.       C.V. %  Lower <-Limit-> Upper

  1) kel                    1.4503      0.802       55.      0.00       10.    
  2) k12                   0.98774      0.522       53.      0.00       10.    
  3) k21                   0.17645      0.286      0.16E+03  0.00       10.    
  4) k13                   0.63700      0.615       97.      0.00       10.    
  5) k31                    1.1047       1.81      0.16E+03  0.00       10.    
  6) V1                     11.569       4.45       38.       1.0       50.    

 AIC =   -8.85493     Final WSS =   0.150183    
               R-squared =   0.9597     Correlation Coeff =   0.9954    

These results represent output from Boomer. Major item is the large values for C.V. %, i.e. from 38 to 160% [Note 0.16E+03 = 160]. We would prefer results below 20% This suggests that there are two many parameters. A smaller model should give a better result.

S.D. results are relative to Value and thus have no particular importance, C.V.% is the term to consider. AIC and WSS don't mean anything alone but need to be compared with different models (same weighting scheme) R-squared and Correlation Coeff are both close to 1 which is good but means that the model (too big) fit the data.

Q 3.3 Comment on the following result:

Plot of Std Wtd Residuals (X)         Plot of Std Wtd  Residuals (X)
   versus time for [Indomethacin]        versus calcd Cp(i) for [Indomethacin] 

    1.824                                  1.824    
 | X                                     |                                X    
 |                                       |                                     
 |                                       |                                     
 |  X                                    |                             X       
 |   X                                   |                           X         
 |                                    X  |X                                    
 |                           X           |  X                                  
 |                                       |                                     
 |                                       |                                     
 |                                       |                                     
 |    X                                  |                        X            
 0X====================================  0====================================X
 |                                       |                                     
 |     X   X        X   X                |    X  X        X     X              
 |                                       |                                     
 |                                       |                                     
 |             X                         |           X                         
 |                                       |                                     
  -0.9630                                -0.9630    
     0.00       <-->             8.0         0.32E-01   <-->            0.52    

These results represent output from Boomer. Weighted residual versus Time and calcd Cp. Note a pronounced 'U' or 'V' shape in both plots. This suggests that the model may be too small, requiring possibly two more parameters.

Adjusting the initial parameter values alone is not enough, the program would just go back to same value Model, i.e. the number of parameters, needs to be changed.

Q 3.4 Comment on the following result:

These results represent output from SAAM II. The plot of Cp and s1 versus time looks OK. That is there is a good correspondance between the observed and calculated data. There is a problem however, with the s1-wres versus time plot, the dashed line. Here we have large deviations, above and below zero, on the left hand side of the plot with much smaller deviations, above and below zero, on the right. This is the 'funnel' shape that can be caused by an inappropriate weighting scheme. Another weighting scheme should be considered.

Q 3.5 Describe the following window:

This window represent input from WinNonlin. This is an input window for WinNonlin were the user can enter initial values for the model parameters and also specify upper and lower limits for these parameters. The user can specify that the program provide initial values and/or limit values. Here the user is specifying both.